ArticlePDF Available

Rockburst prediction and prevention in underground space excavation

July 2023
Underground Space

July 2023

DOI:10.1016/j.undsp.2023.05.009

License
CC BY-NC-ND 4.0

Authors:

Jian Zhou

Central South University

Yulin Zhang

Central South University

Chuanqi Li

Université Grenoble Alpes

Show all 5 authorsHide

The technical challenges associated with deep underground space activities have become increasingly significant. Among these challenges, one major concern is the assessment of rockburst risks and the instability of rock masses. Extensive research has been conducted by numerous scholars to mitigate the risks and prevent occurrences of rockburst through various assessment methods. Rockburst incidents commonly occur during the excavation of hard rock in underground environments, posing severe threats to personnel safety, equipment integrity, and operational continuity. Thus, it is crucial to systematically document real cases of rockburst, allowing for a comprehensive understanding of the underlying mechanisms and triggering conditions. This understanding will contribute to the advancement of rockburst prediction and prevention methods. Proper selection of an appropriate rockburst assessment method is a fundamental aspect in underground operations. However, there is a limited number of studies that summarize and compare different prediction and prevention methods of rockburst. This paper aims to address this gap by analyzing global trends using CiteSpace software since 1990. It discusses rockburst classification and characteristics, comprehensively reviews research findings related to rockburst prediction, including empirical, simulation, mathematical modeling, and microseismic monitoring methods. Additionally, the paper presents a compilation of current rockburst prevention measures. Notably, the paper emphasizes the significance of control strategies, which provide key insights into the effective utilization of stored energy within rock. Finally, the paper concludes by suggesting six directions for implementing intelligent management techniques to mitigate hazards during underground operations and reduce the probability of rockburst incidents.

Content uploaded by Yulin Zhang

Content may be subject to copyright.

Research paper

Rockburst prediction and prevention in underground space excavation

Jian Zhou

a,⇑

, Yulin Zhang

, Chuanqi Li

, Haini He

, Xibing Li

a,⇑

School of Resources and Safety Engineering, Central South University, Changsha 410083, China

Laboratory 3SR, CNRS UMR 5521, Grenoble Alpes University, Grenoble 38000, France

Received 30 November 2022; received in revised form 10 May 2023; accepted 23 May 2023

Available online 22 July 2023

Abstract

The technical challenges associated with deep underground space activities have become increasingly signiﬁcant. Among these chal-

lenges, one major concern is the assessment of rockburst risks and the instability of rock masses. Extensive research has been conducted

by numerous scholars to mitigate the risks and prevent occurrences of rockburst through various assessment methods. Rockburst inci-

dents commonly occur during the excavation of hard rock in underground environments, posing severe threats to personnel safety,

equipment integrity, and operational continuity. Thus, it is crucial to systematically document real cases of rockburst, allowing for a

comprehensive understanding of the underlying mechanisms and triggering conditions. This understanding will contribute to the

advancement of rockburst prediction and prevention methods. Proper selection of an appropriate rockburst assessment method is a fun-

damental aspect in underground operations. However, there is a limited number of studies that summarize and compare diﬀerent pre-

diction and prevention methods of rockburst. This paper aims to address this gap by analyzing global trends using CiteSpace software

since 1990. It discusses rockburst classiﬁcation and characteristics, comprehensively reviews research ﬁndings related to rockburst pre-

diction, including empirical, simulation, mathematical modeling, and microseismic monitoring methods. Additionally, the paper presents

a compilation of current rockburst prevention measures. Notably, the paper emphasizes the signiﬁcance of control strategies, which pro-

vide key insights into the eﬀective utilization of stored energy within rock. Finally, the paper concludes by suggesting six directions for

implementing intelligent management techniques to mitigate hazards during underground operations and reduce the probability of rock-

burst incidents.

Keywords: Rockburst; Underground space; Scientometric analysis; Characteristic analysis; Rockburst prediction; Rockburst prevention

1 Introduction

Rockburst is a signiﬁcant and hazardous phenomenon

that can occur during underground projects involving hard

and brittle rock. It is characterized by its destructiveness,

suddenness, and complexity (Feng et al., 2012; Zhao

et al., 2021; Meng et al., 2021; Askaripour et al., 2022).

The literature reveals that rockburst induced by deep min-

ing can be attributed to various factors, including geologi-

cal factors (such as rock formations, folds, fractures, and

the nature of the ore body), tectonic factors, excavation

factors, and mining design factors (such as excavation

sequence, roadway shape, and pillar design) (Chen &

Zhou, 2023; Jiang et al., 2020; Liu et al., 2020; Wang

et al., 2019a; Yang et al., 2019; Zhou et al., 2021a; Zhou

et al., 2018; Zhu et al., 2018). However, in excavation areas

where the in-situ stresses are suﬃciently high and certain

cracks exist, the sudden release of elastic energy stored

within the rock mass can rapidly trigger a rockburst, lead-

ing to phenomena such as shock, spalling, and zonal disin-

tegration (Cai, 2013;Sun et al., 2021). Rockburst poses a

substantial threat to the safety of workers, equipment,

and subsurface structures and galleries. Suorineni et al.

(2014) describes rockburst as a profound challenge in deep

underground engineering, often referred to as a ‘‘cancer”

due to its detrimental eﬀects.

https://doi.org/10.1016/j.undsp.2023.05.009

2467-9674/Ó2023 Tongji University. Publishing Services by Elsevier B.V. on behalf of KeAi Communications Co. Ltd.

This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

⇑

Corresponding authors.

E-mail addresses: j.zhou@csu.edu.cn (J. Zhou), xbli@csu.edu.cn (X.

Li).

www.keaipublishing.com/undsp

Available online at www.sciencedirect.com

ScienceDirect

Underground Space 14 (2024) 70–98

Given the increasing depths of civil engineering and

mining activities in underground projects, seismic activity

resulting from excavation and rockburst issues has become

inevitable and cannot be eﬀectively prevented (Li et al.,

2019). Although rockburst occurrences are likely to take

place in early underground projects, the ﬁrst oﬃcially doc-

umented case of rockburst was reported in a tin mine in

England (Zhou et al., 2012). Since then, numerous inci-

dents of varying intensity and impact have been reported

in projects such as mines, tunnels, shafts, underground lab-

oratories, and hydroelectric facilities in countries such as

China, the United States, South Africa, Australia, and

Canada, coinciding with the deepening of operating depths

(Li et al., 2017b). Figure 1 provides an overview of excava-

tion depths in various regions, as documented by Li et al.

(2017b), Rehman et al. (2021),Xi et al. (2017), and

Hudson and Feng (2015). Notably, over 20 mines in

Canada have encountered rockburst accidents. In the Uni-

ted States, between 1936 and 1993, a total of 172 recorded

rockburst events resulted in 78 fatalities and 158 injuries

(Mark, 2016; Pu et al., 2019). Despite a decline in mining

activities, Germany reported more than 40 cases of rock-

burst from 1983 to 2007, all of which caused signiﬁcant

injuries and fatalities. In China, numerous metal mines

have faced challenging dynamic hazards such as mine tre-

mors, roof collapse, high-intensity rockburst incidents,

and extensive destabilization of goaf. These hazards are

notoriously diﬃcult to predict and eﬀectively prevent (Du

et al., 2022; Li & Gu, 2002; Pu et al, 2019; Zhou et al.,

2020, 2015b, 2016b). Hence, prioritizing prevention and

the control of these problems is of utmost importance in

underground excavation.

Based on the information presented above, rockburst

incidents commonly arise in areas characterized by high

in-situ stress and brittle hard rock formations. The man-

agement of high magnitudes of in-situ stress represents a

primary challenge when undertaking underground projects

in deep environments. Failure to adopt appropriate exca-

vation techniques in high-stress environments will inevita-

bly lead to large-scale engineering disasters. Moreover,

the occurrence of rockburst directly disrupts project oper-

ations and production, thereby signiﬁcantly impacting the

economic beneﬁts of the company. Therefore, it is crucial

to have the ability to eﬀectively summarize typical rock-

burst cases and visually review the research ﬁeld, as this

provides a factual foundation for in-depth investigations

into the mechanisms and classiﬁcations of rockburst. The

visual analysis process is illustrated in Fig. 2.

Due to the complex nature of rockburst, eﬀective predic-

tion and prevention of rockburst poses signiﬁcant chal-

lenges (Zhou et al., 2010). The causes of rockburst in

hard rock mines are multifaceted, involving unclear mech-

anisms, uncertain causes, and inaccurate location determi-

nation, leading to severe consequences. Figure 3 provides a

Fig. 1. Excavation depths for typical underground project.

J. Zhou et al. / Underground Space 14 (2024) 70–98 71

detailed illustration of these speciﬁc factors. However,

given the multitude of inﬂuential factors contributing to

rockburst, establishing a general mechanical theory of

rockburst mechanism, particularly regarding fault slip

bursts, remains elusive and necessitates further research.

These challenges are inherent to the deep engineering envi-

ronment. Therefore, it is crucial to carefully consider the

selection of prediction and prevention technologies based

on the engineering geological conditions of the rock mass.

The remaining sections of this paper are structured as

follows. In Section 2, the global collaborations, hotspots,

and trends in the ﬁeld of rockburst are analyzed using

CiteSpace software (Chen, 2006).Section 3 provides a sum-

mary of typical cases of rockburst in underground projects

and elucidates the corresponding rockburst characteristics.

Traditional methods of rockburst prediction and microseis-

mic monitoring techniques are reviewed in Section 4.Sec-

tion 5 presents the underlying control strategies and

support methods for mitigating rockburst, emphasizing

the framework for safe and eﬃcient excavation. The future

directions of excavation in underground space are dis-

cussed in Section 6. Finally, in Section 7, comprehensive

conclusions are drawn based on the extensive review con-

ducted in this study.

2 Scientometric review on rockburst

The occurrence of rockburst hazards during under-

ground excavation represents a signiﬁcant scientiﬁc chal-

lenge that demands attention. These hazards arise from

the sudden and intense release of accumulated high-

magnitude strain energy in deep underground engineering.

However, due to an unclear understanding of the progres-

sive failure process, eﬀectively addressing these challenges,

including precise prediction and eﬀective prevention of

rockburst, remains highly diﬃcult. Consequently, this issue

has become a universal concern worldwide, as evidenced by

documented cases in countries such as South Africa (Leger,

1991), Canada (Simser, 2019), the United States (Blake &

Hedley, 2003), China (Zhou et al., 2012; Zhang et al.,

2011), and others (Barton et al., 1974). Rockburst incidents

occur in various contexts, including the construction and

resource extraction of deep mines, hydroelectric tunnels

and chambers, nuclear waste geological treatment, and

deep physical underground laboratories. These incidents

lead to signiﬁcant economic losses, casualties, and schedule

delays. In general, there exists a positive linear correlation

between the number and severity of rockburst events and

the depth of excavation.

Fig. 2. Research methodology of visual review for rockburst research.

72 J. Zhou et al. / Underground Space 14 (2024) 70–98

To date, there has been a limited number of papers pro-

viding a comprehensive analysis of the overall research sta-

tus of rockburst. As scholars increasingly focus their

attention on rockburst research, it is necessary to conduct

a systematic analysis of the current state of this ﬁeld. Such

analysis can help identify the existing knowledge gaps and

explore potential research directions. Various software

options are available today for visual analysis, each with

its own advantages and limitations. Among these options,

CiteSpace stands out due to its comprehensive features, its

focus on analyzing the evolutionary path of the research

ﬁeld, and its ability to highlight emerging and cutting-

edge topics. These features align well with our goal of con-

ducting a thorough analysis. For this study, CiteSpace soft-

ware is employed to perform a scientometric analysis of

2,645 English literature articles from the Web of Science

Core Collection (WOSCC) database. The literature search

in WOSCC includes the following criteria: subject = (rock

burst or rock burst), language restricted to English, and the

time span is set from 1990 to 2022.

2.1 Global cooperation analysis

Given the numerous research eﬀorts devoted to the

ﬁeld of rockburst, it is imperative to display the spatial

distribution of articles in a quantitative manner through

co-country analysis. Figure 4(a) illustrates the leading

countries that have made substantial contributions to

rockburst research. The top 5 countries with the highest

number of publications are China (1584 articles), the US

(171 articles), Australia (153 articles), Canada (130 arti-

cles), and Russia (122 articles), respectively. Notably,

these countries initiated research on rockburst in the con-

text of underground excavation at an early stage and

have made signiﬁcant contributions to the ﬁeld. Further-

more, the innovative research results in rockburst mainly

come from these countries, which indicates their domi-

nant position in the target domain. It is worth paying

special attention to China, Australia, Canada, and Rus-

sia, as these countries have a correlated signiﬁcant role

in the ﬁeld. As a developing country, China has realized

the importance of disaster management and conducted

in-depth research on rockburst over the last two decades.

Jiang et al. (2010) introduced a novel energy index for

simulating rockburst conditions during tunneling activi-

ties at the Jinping II hydropower plant. This index was

computed by capturing the peak and trough values of

elastic strain energy intensity before and after brittle

damage occurred. In a study by Feng et al. (2004),

indoor tests were conducted to analyze the evolution

Fig. 3. Status of underground space excavation.

J. Zhou et al. / Underground Space 14 (2024) 70–98 73

mechanism of sandstone damage. Real-time CT tech-

niques were employed to subject sandstone specimens

to triaxial loading with chemical corrosion. Computed

tomography images and CT values of the sandstone

specimens were obtained during compression, microfrac-

ture, and expansion until failure at diﬀerent loading

levels. Tang (1997) proposed a rock failure process anal-

ysis (RFPA) method based on ﬁnite element theory to

simulate the process of rock damage. Subsequently,

Wang et al. (2006) investigated the rockburst model of

ore pillars using the RFPA 2D program. Zhu et al.

(2010) developed a numerical model for the dynamic dis-

turbance of rockburst in underground excavated cavities.

Canada and the United States have also made signiﬁcant

contributions to the ﬁeld of rockburst research over a

considerable period. Kidybinski (1981) proposed a series

of rockburst propensity indices for coal mines. Kaiser

et al. (1992) published rockburst assessment procedures

for underground mining engineering. The co-occurrence

analysis of countries reveals a close collaboration

between researchers and research institutions in promi-

nent mining countries. Cook (1965, 1983) provided valu-

able insights and observations on the origin of rockburst

in South Africa and the stability of rockburst events.

Co-institution analysis enables the identiﬁcation of core

research strengths in the ﬁeld and provides a scientiﬁc eval-

uation of the academic impact of these institutions. Figure 4

(a) illustrates the top 9 institutions based on publication fre-

quency. The institution with the highest publication fre-

quency is China University of Mining and Technology,

followed by Shandong University and the Chinese Acad-

emy of Sciences. These higher education institutions spe-

cializing in the mining industry have emerged as major

contributors to current rockburst research. In Canada,

prominent institutions include McGill University (24 arti-

cles) and the University of Toronto (21 articles). Addition-

ally, several internationally renowned scientiﬁc institutions,

such as the Russian Academy of Sciences (86 articles) and

Monash University (20 articles), have made signiﬁcant con-

tributions. Notably, the top 3 publishers in terms of institu-

tional aﬃliation are all Chinese institutions, indicating the

prominent position and high level of attention given to

rockburst research within the Chinese academic commu-

nity. While global communication has become more acces-

sible with the rapid expansion of the Internet, institutional

cooperation remains largely dominated by institutions

and disciplines with similar interests, resulting in less inter-

disciplinary and cross-disciplinary collaboration.

Fig. 4. (a) Research countries and issuing institutions in the ﬁeld of rockburst and percentage of publications, and (b) annual volume of documents.

74 J. Zhou et al. / Underground Space 14 (2024) 70–98

Figure 4(b) illustrates the publication count pertaining

to the domain of rockburst. Evidently, there exists a dis-

cernible upward trend in the number of published papers

within this ﬁeld, accompanied by a J-shaped growth in cita-

tion frequency. This observation signiﬁes the escalating

global interest in the topic of rockburst.

2.2 Research frontier analysis

Keywords serve as the core and summary of an article,

providing a concise description and condensation of the

research content. Thus, keywords with high frequency

can eﬀectively identify the hot topics within the research

domain of rockburst. Table 1 presents the detailed results

of burst keywords, with the highest occurrence intensity

observed for ‘rockburst’ at 8.93. This indicates that rock-

burst has been a persistent hazard in underground excava-

tion from 1999 to 2010. The seismic intensity of rockburst

has had a lasting and widespread impact since its initial

occurrence in 1999. Over the past decade, keywords such

as ‘‘electromagnetic radiation,”‘‘numerical simulation,”

‘‘mathematical models,”and ‘‘microseismic monitoring”

have been employed in rockburst prediction. Researchers

have also shown great interest in studying the ‘‘rockburst

mechanism”and ‘‘energy release”to elucidate the pro-

cesses involved in pillar burst and fault-slip burst move-

ment, thereby gaining a deeper understanding of the

occurrence mechanism and its impact. The burst strength

of keywords such as ‘rock support’ and ‘rockburst disaster’

reaching 3.91 and 4.07, respectively, indicates that security

concerns related to deep mining have been a signiﬁcant

focus for various countries. Overall, the diverse range of

these keywords reﬂects the extensive topics of interest in

this ﬁeld over a speciﬁc time span. The keyword time-line

graph (Fig. 5) provides a general overview of rockburst

studies. Early rockburst research primarily focused on deep

excavation, mine seismicity, and rockburst damage.

Experts employed rock mechanics engineering knowledge

and conducted physical model experiments to explore the

damage process of rockburst, leading to the proposal of

prediction methods such as acoustic emission. Subse-

quently, various prediction methods, including stress

strength, energy release, numerical simulation, machine

learning, and microseismic monitoring, have been

expanded upon to enhance our understanding of failure

mechanisms and facilitate preventive measures. Notably,

rockburst research is progressing in integration with the

ﬁeld of artiﬁcial intelligence. For example, Adoko et al.

(2013) employed fuzzy inference system and adaptive

neuro-fuzzy inference system to classify 174 rockburst

cases in China. Similarly, Zhou et al. (2016a) used ten

supervised learning methods to classify 246 rockburst

events. A comprehensive review of recent advances in rock-

burst prediction was conducted by Zhou et al. (2018),

which highlights the application of machine learning tech-

niques, among other methods, to improve the accuracy of

prediction models

3 Rockburst statistical analysis

The objective of rockburst classiﬁcation is to identify the

crucial point for statistical analysis of rockburst occur-

rences in underground excavations. Before conducting the

statistical analysis, it is essential to precisely establish the

deﬁnitions and classiﬁcation of rockburst and understand

their underlying mechanisms. The challenges associated

with rockburst discussed earlier are considered some of

the most diﬃcult in the ﬁeld. Numerous researchers have

conducted scientiﬁc analyses focusing on the deep rock

environment and stress conditions throughout the excava-

tion process (Meng et al., 2012, 2013).

Regarding rockburst phenomena during excavation,

Oliver et al. (2022) identiﬁed three rockburst mechanisms

in Ontario’s Canadian mines (Fig. 6). Strainburst-induced

Table 1

Top 15 keywords with the strongest citation bursts.

Keywords Strength Begin End 1999–2021

Earthquake 8.93 1999 2010

Electromagnetic radiation 4.04 2000 2004

Fracture 5.39 2004 2011

Growth 4.72 2005 2015

Numerical simulation 5.07 2006 2016

Rockburst 27.73 2008 2015

Rock mechanics 6.06 2008 2015

Hard roof 5.01 2012 2017

Deep mining 4.31 2012 2017

Rockburst hazard 4.07 2012 2014

Rock support 3.91 2012 2013

Stress–strain state 3.83 2012 2018

Energy release 3.64 2013 2016

Event 3.44 2013 2017

Longwall mining 3.56 ‘2014 2018

Notes: The blue section shows the time span from 1999 to 2021. The red part indicates the time span from the occurrence to the termination of the burst

keywords.

J. Zhou et al. / Underground Space 14 (2024) 70–98 75

stress is concentrated behind the excavation face, and the

region between this stressed zone and the free face can

become unstable, resulting in sudden failure and ejection

of rock material into the opening. In pillar burst, stress is

concentrated in the core of the pillar, leading to overstress-

ing, sudden failure, and material ejection into surrounding

openings. For fault-slip burst, changes in the stress acting

on the locking point can alter the shear stress (s) and

clamping normal stress (r

). Deng and Gu (2018) treated

rockburst as a structural instability problem and classiﬁed

it into three categories based on the magnitude of dynamic

stimulation force: intrinsic rockburst, triggered rockburst,

and induced rockburst. Keneti and Sainsbury (2018) char-

acterized 26 rockburst events and concluded that the causes

of rockburst are related to rock loading conditions and

rock properties (Fig. 7). Russenes (1974) assessed rock-

burst intensity based on changes in the environmental state

(e.g., damage noise, shape, and characteristics) at the time

of occurrence, resulting in a classiﬁcation into four grades.

The Canadian Rockburst Research Project employed the

depth and geometry of rock damage zones to classify

rockburst damage as minor, moderate, and severe

Fig. 5. Timeline chart for rockburst keywords. (#0: acoustic emission; #1: coal burst; #2: rock burst; #3: microseismic monitoring; #4: rock bursts; #5:

depth; #6: failure; #7: rock discontinuity).

Fig. 6. Three typical rockburst types and mechanisms.

76 J. Zhou et al. / Underground Space 14 (2024) 70–98

(Kaiser et al., 1996; Kaiser & Cai, 2012). In China, Chen

et al. (2015) proposed a quantitative method for evaluating

rockburst classiﬁcation based on the energy dissipated by

rockburst radiation and the extent of surrounding rock

damage.

In recent years, there has been an increase in the con-

struction and excavation of deep and long underground

space projects. Extensive literature highlights the chal-

lenges in accurately predicting and preventing hazards such

as high-energy rockburst and mine seismicity in these deep

environments. Rockburst is one of the most prominent and

complex issues encountered in such projects. Therefore, it

is crucial to compile and analyze rockburst data from var-

ious sources to provide reference experiences for evaluation

and prevention, as well as theoretical support for in-depth

dynamic disaster research. To assess the occurrence and

severity of rockburst, Zhou (2015a) compiled 102 cases of

rockburst in hard rock mines in China (Fig. 8(a)). They

developed a classiﬁcation model utilizing variables such

as maximum tangential stress around the excavation

(MTS or r

), uniaxial compressive strength of rock (UCS

or r

), uniaxial tensile strength of rock (UTS or r

), stress

concentration factor (SCF or r

), rock brittleness index

(B = r

), and elastic strain energy storage index (W

The stochastic gradient boosting method was used to ana-

lyze the data and determine the relative importance of each

variable. The results (Fig. 8(c)) indicate that the Wet index

is the most sensitive in predicting rockburst intensity, fol-

lowed closely by MTS. SCF and B are more sensitive than

UTS and UCS. Furthermore, Heal (2010) collected 254

databases of rockburst events in Canada and Australia

(Fig. 8(b)), considering variables such as SCF, ground

support capacity (GSSC), geological structures (GS),

excavation structures (ES), and peak particle velocity

(PPV) at the site. The same analysis method was employed

to identify the main controlling factors of rockburst occur-

rence, as depicted in Fig. 8(d). PPV is the most sensitive

index in assessing the tendency for rockburst damage, fol-

lowed by ES. SCF is more sensitive than GSSC and GS.

These research ﬁndings demonstrate the feasibility and

practicality of artiﬁcial intelligence-driven methods in rock-

burst studies. By gaining a comprehensive understanding

of the primary controlling variables of rockburst occur-

rence, the risk can be eﬀectively mitigated in the design

of underground projects.

3.1 Disaster statistics analysis

Drawing systematic conclusions from a limited sample

size or a small number of mine rockburst instances poses

signiﬁcant challenges. To establish the general patterns of

rockburst, statistical analysis was conducted on several

typical cases of underground hard rock engineering (refer

to Appendix A). The analysis considered factors such as

the maximum burial depth, rock characteristics, and

observed rockburst phenomena. The statistical results indi-

cate that rockburst events occurring below 1000 m depth

are predominantly of weak to moderate intensity, with a

few cases classiﬁed as severe. The macro manifestations

of rockburst include the rattling sound of rocks, rock ejec-

tion, tunnel fragmentation, roof collapse, and surrounding

rock failure. In the initial stages, rockburst incidents above

1000 m depth are typically minor to moderate in severity.

However, as excavation progresses and the construction

area expands, the intensity of rockburst tends to increase

and may reach high or severe levels in the later stages. At

Fig. 7. Factors contributing to rockburst.

J. Zhou et al. / Underground Space 14 (2024) 70–98 77

this point, rockburst events are accompanied by loud blast-

ing sounds, noticeable surface tremors, intermittent crack-

ing sounds, and occasional dull popping sounds. These

ﬁndings lead to the conclusion that, irrespective of the type

of hard rock and burial depth, rockburst incidents of vary-

ing severity can occur in practical engineering scenarios.

3.2 Rockburst characteristics analysis

This review aims to collect and compile information on

the characteristics and causes of rockburst phenomena in

typical hard rock projects worldwide. Additionally, it pro-

vides a grading of rockburst severity, serving as a founda-

tion and guidance for the development of subsequent

disaster management measures to mitigate rockburst risks

in deep excavations. Based on the experience gained from

treating typical rockburst projects, this review extensively

describes the classiﬁcation and characteristics of rockburst,

as presented in Table 2. The ratio between the maximum

shear stress and the uniaxial compressive strength of the

rock serves as the fundamental criterion for determining

the level of rockburst. Subsequently, various factors

including movement characteristics, sound characteristics,

phenomenological characteristics, incision shape character-

istics, time characteristics, and the degree of damage are

employed to identify appropriate prevention strategies.

4 Rockburst prediction

Before delving into the discussion of suitable prevention

measures for diﬀerent types of rockburst, it is essential to

provide a comprehensive overview of rockburst prediction.

The prediction of rockburst events is crucial in eﬀectively

preventing and controlling rock damage in underground

Fig. 8. Data visualization of rockburst and feature importance in underground projects. Dataset visualization: (a) Zhou (2015a) rockburst cases, and (b)

Heal (2010) rockburst cases; Feature importance: (c) Zhou (2015a) rockburst cases, and (d) Heal (2010) rockburst cases.

78 J. Zhou et al. / Underground Space 14 (2024) 70–98

excavation and mining projects, considering the complex

nature of this phenomenon. Since the 1960s, extensive

research has been conducted on various techniques for

forecasting rockburst occurrences. Researchers have made

eﬀorts to develop, reﬁne, and calibrate prediction methods

using diverse approaches. The prediction methods for

rockburst can be broadly categorized into ﬁve aspects:

empirical methods, simulation techniques, mathematical

algorithms, rockburst charts, and monitoring technologies,

as depicted in Fig. 9.Figure 10 presents a reﬁned and

adapted representation of the major advancements in

empirical, numerical, statistical, and intelligent classiﬁca-

tion methods for rockburst evaluation from 1965 to 2021,

based on the temporal analysis of various classiﬁcation

methods summarized by Zhou et al. (2018).

4.1 Empirical criteria method

In the initial stages of engineering design and construc-

tion, empirical criteria are commonly employed to assess

the quality of rock mass and evaluate the risk of rockburst,

providing practical guidance for project design and sup-

port. The empirical criteria encompass both single-

indicator and multi-indicator integrated criteria, utilizing

a range of indicators to determine the classiﬁcation of rock-

burst intensity based on the analysis of physical tests and

mechanical factors in engineering case studies. Building

upon the main achievements and theoretical methods of

rockburst prediction, numerous single index criteria have

been proposed to assess rockburst risks. These criteria

include the Torchaninov criterion (Turchaninov et al.,

1972), Barton criterion (Barton et al., 1974), Tao Zhenyu

criterion (Tao, 1988), Russense criterion (Russenes,

1974), Hoek criterion (Brown & Hoek, 1980), Wet’s strain

energy storage index (Kidybin

´ski, 1981), Excavation vul-

nerability potential (Heal et al., 2006), Potential stress fail-

ure (Mitri, 2007), Erlang mountain method (Wang et al.,

1999), Hou criterion (Hou and Wang, 1989), Composite

index criterion (Tan et al., 1991), Rockburst vulnerability

index (Qiu et al., 2011), maximum stored elastic strain

energy index ES (Guo et al., 2011), improved brittleness

index BIM (Zhang et al., 2017) and rockburst danger index

(Dou et al., 2012), and others. The empirical criteria serve

as important qualitative tools for gaining a comprehensive

understanding of rock engineering issues, particularly when

reliable object label information is available. The results

obtained from these criteria for assessing rockburst poten-

tial can provide early risk warnings for underground exca-

vation projects. The various approaches and methods

associated with empirical criteria will be explained and

compared in detail in the subsequent discussion.

In the context of predicting rockburst events, single-

indicator criteria, primarily stress and energy approaches,

are widely recognized as the primary indicators for classify-

ing the intensity of rockburst. The former focuses on

geomechanical characteristics and induced stress levels,

while the latter considers the stored elastic energy within

Table 2

Rockburst classiﬁcation and characteristics (Russenes, 1974; Chen et al., 2015; Zhou et al., 2018; Ke, 2021).

Classiﬁcation No rockburst Light rockburst Medium rockburst Heavy rockburst Serious rockburst

Movement features No stability

problems

Wall caving Wall caving, spalling Spalling and even ejection New fracture face

Failure type The crack

occurred inside

rock mass

slight spalling and slabbing in the

surface of the surrounding rock mass;

the rock mass was slightly ejected

severe spalling and slabbing of the

surrounding rock mass; the rock

mass was obviously ejected

A great deal of rock mass

was suddenly ejected; the

failure range was extensive

A large block of rock mass was suddenly

ejected with intensive seismicity and the

stability of the whole carve was seriously

aﬀected.

Sound features No noises Light noises Strong cracking noises Rock noises of gun shot

strength

Loud sounds like dull thunder

Depth of failure (D

)<0:5m 0:5:0m 1:02:0m >3:0m

Size of ejected

fragment

<10 cm 10–30 cm 30–80 cm 80–150 cm >150 cm

Damage degree No support system

and construction

are aﬀected

Neither the support system nor

construction are damaged

The shotcrete lining could be

damaged among rock bolts,

construction is slightly aﬀected

Support system is

destroyed and construction

are aﬀected.

Support system is seriously destroyed, and

construction is seriously aﬀected.

Prevention methods No support system Appropriate safety measures rock bolt and shotcrete lining are

constructed in time

Set up inverted arch if

necessary

Use special means to prevent

J. Zhou et al. / Underground Space 14 (2024) 70–98 79

the rock mass (Cook, 1963). Among stress estimation

approaches, the rock brittleness coeﬃcient and tangential

stress criterion have demonstrated reliable results in fore-

casting rockburst intensity. These criteria rely on simple

assessments of rock mass properties, which can be deter-

mined through practical experiments and provide realistic

values. Energy approaches, on the other hand, are more

commonly used to evaluate the severity of rockburst. These

methods oﬀer greater insights into the causes of rockburst

by comparing the results with seismicity data collected in

burst-prone areas. Another reason for the adoption of

energy methods is that the energy evolution of the rock

mass, including storage, release, and dissipation, is directly

associated with rockburst occurrence. Hence, compared to

other criteria, energy-based approaches are more likely to

indicate the tendency for rockburst. Multi-index indicators,

such as the ﬁve factors criterion, have gained popularity as

they consider multiple indicators and signiﬁcant rock prop-

erties, enabling a more comprehensive assessment of rock-

burst intensity. The empirical criteria and severity

classiﬁcation methods employed in the assessment of rock-

burst disasters in underground excavation projects are

highly regarded for their practicality and feasibility. Pre-

dicting rockburst potential and classifying rockburst sever-

ity are two fundamental aspects in evaluating rockburst

approaches. The severity classiﬁcation follows four levels:

none, weak, moderate, and severe (Russenes, 1974). In

most cases, the thresholds for intensity levels are deter-

mined based on analytical and statistical characteristics

of the region where rockburst events have been recorded,

or through engineering experience (Zhou et al., 2012;

Feng et al., 2013). However, the classiﬁcation of rockburst

intensity may vary across diﬀerent projects, and using the

same threshold as evaluation criteria presents signiﬁcant

drawbacks. For instance, Kidybin

´ski (1981) suggested that

a Wet index greater than 5 indicates the potential for severe

rockburst, while Singh (1987) proposed that Wet < 10

corresponds to no or weak rockburst intensity, based on

experimental ﬁndings for hard rock samples from the Sud-

bury area in Ontario, Canada. The variety of rockburst cri-

teria makes it diﬃcult for frontline engineers to accurately

determine the severity of rockburst disasters (Zhou et al.,

2018).

4.2 Simulation technology

Various types of rockburst occurrences are inﬂuenced by

a multitude of geomechanical parameters, which have a

signiﬁcant impact on the underlying mechanisms and

observable characteristics of rockburst. In addition to these

contributing factors, the magnitude and orientation of

in situ stresses, dynamic disturbances, excavation geome-

try, and advancement rate also play a crucial role in deter-

mining the severity, shape, and location of rockburst

events. As mentioned earlier, empirical criteria for rock-

burst primarily include energy criteria and stress criteria.

These methods ﬁnd extensive application in diﬀerent

aspects of rock engineering, such as classifying the severity

of rockburst and serving as input parameters for numerical

simulation methods. According to Gu (2013), numerical

methods and physical experiments have become essential

tools for tackling scientiﬁc problems that were previously

diﬃcult to address when there is ample labeled information

available to engineers and researchers. Numerical simula-

tion experiments constitute the majority of research eﬀorts

related to rockburst prediction. Zubelewicz and Mroz

(1983) utilized the ﬁnite element method (FEM) and kinetic

methods to simulate the destabilization damage of rock

masses and investigate the dynamic behavior of rockburst.

Bardet (1989) demonstrated the eﬀectiveness of FEM in

predicting surface instability and proposed that FEM can

be employed to analyze rockburst problems by considering

them as surface instability issues, as evidenced by the eigen-

value method. Mu

¨ller (1991) compared the performance of

Fig. 9. Major methods of rockburst prediction.

80 J. Zhou et al. / Underground Space 14 (2024) 70–98

FEM-based and ﬁnite diﬀerence method (FDM)-based

approaches in simulating rockburst, and it was found that

FDM is the more appropriate method. It is evident that

dynamic damage plays a signiﬁcant role in the formation

and occurrence process of rockburst. However, capturing

the true development processes of instantaneous behavior

using numerical simulations can be challenging. To assess

rockburst hazards, Tajdus

´(1997) developed a three-

dimensional numerical stress ﬁeld calculation model based

on the ratio of induced stress to original stress conditions.

To investigate the eﬀect of non-uniformity on rock frac-

ture, Ishida et al. (2009) proposed a simulation of uniaxial

compression tests using a discrete element method (DEM)

program. Additionally, many experts have conducted var-

ious physical model tests. Notably, several researchers in

China (Xu et al., 2002; Li et al., 2001; Yang, 1993) have uti-

lized similar materials in their experiments to model rock-

burst during tunnel construction. Lu et al. (2008)

proposed an analogous materials model to simulate stress

wave propagation behavior and the damage process of

hard rock. Du et al. (2016) studied the damage behavior

of granite, red sandstone, and cement mortar using an

advanced test system that combines genuine triaxial static

stress and local dynamic disturbance. The test results

revealed that sandstone and granite experienced slab spal-

ling in true triaxial unloading tests, while no signiﬁcant

fractures occurred under low amplitude local dynamic

loading. However, the degree of rock damage increased

with higher local dynamic loading amplitude, regardless

of the loading direction. In another study, Du et al.

(2022) investigated the mechanical characteristics of

rectangular prismatic rock samples and isolated columns

made of granite, marble, and sandstone with various

height-to-breadth and width-to-thickness ratios under uni-

axial compressive stress.

Future studies will require a signiﬁcant number of engi-

neering instances to be statistically examined, as it is cur-

rently not possible to establish a suitable classiﬁcation

system for rockburst solely through numerical simulation

without an acceptable energy criterion. Although physical

model tests have limitations compared to numerical simu-

lation methods, they still oﬀer certain advantages. For

instance, while it may not be possible to fully simulate

the real engineering environment (including excavation dis-

turbances and methods) or incorporate artiﬁcial boundary

conditions in rockburst simulations, physical model tests

provide a high degree of restoration of the engineering

environment and yield credible data results. As a result,

some scholars continue to employ physical model tests to

address complex issues encountered during the excavation

of hard rock.

4.3 Mathematical modeling method

The precision of rockburst predictions relies heavily on

the quantity and quality of the data used (Qiu & Zhou,

2023; Shi et al., 2010). Mathematical models are commonly

developed using collected input and output data, allowing

them to handle diverse data types. Therefore, it is feasible

to construct suitable mathematical models based on empir-

ical criteria for data-driven prediction of previous rock-

burst cases. Data-driven prediction can be broadly

Fig. 10. Major work on mathematical methods, numerical simulations and empirical criteria. The red and white outer boxes represent the English and

Chinese literature respectively. The diﬀerent color blocks depict author information and methods for diﬀerent research directions.

J. Zhou et al. / Underground Space 14 (2024) 70–98 81

categorized into two groups: uncertainty theory and

machine learning.

In situations where there is insuﬃcient rockburst data,

statistical-based methods or artiﬁcial intelligence tools have

limitations in their application. Considering that the inﬂu-

encing factors of rockburst are stochastic and fuzzy, uncer-

tainty theories have been introduced in rockburst studies

(Zhou et al., 2018). These theories include unascertained

mathematical theory (Jin et al., 2017), fuzzy mathematical

theory (He et al., 2021), catastrophe theory (Qiao et al.,

2021), grey system theory (Pei et al., 2013), cloud models

(Liu et al., 2019; Wang et al., 2020), rough set theory (Jia

et al., 2014), extension theory (Zhang et al., 2020c), attri-

bute mathematical theory (Wen, 2008), interval number

theory (Wang et al., 2019b), and set pair analysis (Chen

et al., 2008), as shown in Fig. 11. Recently, several hybrid

comprehensive evaluation models have been utilized for

rockburst assessment. The multidimensional connected

cloud model proposed by Wang et al. (2020) oﬀers a sim-

pler and more convenient approach for practical applica-

tions compared to the one-dimensional cloud model.

Liang et al. (2019) employed an extended multi-attribute

boundary approximation area comparison method based

on triangular fuzzy numbers to evaluate rockburst tenden-

cies in four critical areas of the Kaiyang phosphate mine.

Liu et al. (2019) developed a rockburst classiﬁcation model

based on rough set and cloud model theory, which was val-

idated using ﬁve rockburst cases. Zhou et al. (2022a) com-

bined four indicators and utilized set pair analysis to

evaluate coal burst liability, creating an improved connec-

tion cloud model (ICCM). The results indicate that the

ICCM exhibits signiﬁcantly superior classiﬁcation index

accuracy and Kappa coeﬃcient of 0.88 and 0.772, respec-

tively, compared to other conventional uncertainty

approaches.

The machine learning approach has several advantages

over other methods, as it does not require prior knowledge

about the relationship between input and output variables,

thereby reducing human intervention. This has made it a

popular choice for researchers in solving nonlinear prob-

lems. Various machine learning algorithms such as support

vector machines (SVM) (Zhou et al., 2012), logistic regres-

sion classiﬁers (Li & Jimenez, 2018), decision trees (DT)

(Shirani & Taheri, 2019), Bayesian networks (N. Li et al.,

2017a;C. Li et al., 2021), random forest (RF) and

gradient-boosting machine (GBM) (Zhou et al., 2016a),

and K-nearest neighbor (KNN) (Afraei et al., 2019) have

been employed for classiﬁcation and prediction of rock-

burst. Ensemble learning is a method of enhancing learning

ability by combining multiple weak classiﬁers (Wang et al.,

2021). Recently, Li et al. (2022a) proposed a deep forest

rockburst prediction model based on a database of 329

actual rockburst cases. The model achieved a ﬁnal test set

accuracy of 90.48% and a performance gain of 12.7%. In

another study, Li et al. (2022b) utilized a Bayesian opti-

mization and synthetic minority oversampling tech-

nique + Tomek Link approach to build a feedforward

neural network model based on 314 rockburst examples.

Similarly, Li et al. (2022c) developed rockburst prediction

methods using combined trees, such as random forest,

extremely randomized tree, adaptive boosting machine,

gradient boosting machine, extreme gradient boosting

machine, light gradient boosting machine, and category

gradient boosting machine, using the same rockburst data-

set. They integrated three voting, bagging, and stacking

procedures with a Bayesian optimization algorithm to

enhance model performance. Zhang et al. (2020b) com-

bined seven classiﬁers and nine data interpolation methods

to solve the rockburst severity classiﬁcation problem. The

results showed a 15.4% improvement in accuracy over the

best individual classiﬁer. Yin et al. (2021) used the ensem-

ble learning stacking idea to predict the rockburst potential

by developing four ensemble learning models: KNN-

Recurrent neural network (RNN), SVM-RNN, Deep neu-

ral network (DNN)-RNN, and KNN-SVM-DNN-RNN.

Zhou et al. (2021b) developed a hybrid model combining

of ﬁreﬂy algorithm and artiﬁcial neural network (ANN)

to deal with various complex relationships of rockburst

in deep hard rock mines.

4.4 Microseismic monitoring technology

Rockburst events involve complex processes of rock

deformation and fracture extension. The utilization of

microseismic monitoring technology allows for the acquisi-

tion and analysis of information regarding the develop-

ment of rock fractures through a microseismic system

(Fig. 12). Figure 12(a) illustrates the operational mode of

a typical microseismic positioning system in metal mines.

Over the years, three major challenges in underground

space excavation have been the signiﬁcant errors in locat-

ing disaster sources, diﬃculties in identifying microseismic

signals, and the lack of accurate warning and protection

mechanisms (Zhu et al., 2021). In addressing these chal-

lenges, Li and Dong (2014) proposed a theoretical system

for microseismic source location without pre-velocimetry

and developed fully automatic software for processing

microseismic data based on wave velocity determination

without pre-measurement, resulting in a signiﬁcant

improvement in localization accuracy. Perol et al. (2018)

conducted an initial exploration of deep learning algo-

rithms for microseismic localization. They employed con-

volutional neural networks to initially localize

microseismic events into six major regions, followed by cal-

culating the seismic wave delay time in underground mines

and identifying the location of the earthquake source. Fig-

ure 12(b) represents the microseismic ﬂow diagram, while

Fig. 12(d) depicts microseismic signal identiﬁcation and

learning (Ma et al., 2015).

Microseismic location systems typically require both

speed and accuracy. However, the pursuit of high precision

often leads to longer computation times. To address this

issue, Zhou et al. (2022b) applied three metaheuristic algo-

rithms to enhance the performance of Cross-correlation

82 J. Zhou et al. / Underground Space 14 (2024) 70–98

stacking and increase computational eﬃciency. The eﬀec-

tiveness of this approach was demonstrated through engi-

neering applications at the Fankou Pb-Zn mine in China.

The virtual ﬁeld optimization method (VFOM), which is

a rapid and precise microseismic locating technique, has

gained popularity within the mining community. It can

accurately locate microseismic sources even in the presence

of signiﬁcant picking errors. However, the VFOM’s objec-

tive function complexity, particularly when multiple sen-

sors are used, may result in longer localization times. To

overcome these challenges, Zhou et al. (2021c) utilized

two heuristic algorithms to improve the location eﬃciency

of the VFOM with minimal accuracy loss and to avoid

convergence to local optimal values.

According to Cai et al. (2007), acoustic emission and

microseismic activity are low-energy seismic events associ-

ated with sudden inelastic deformation, such as the abrupt

movement of existing fractures, generation of new frac-

tures, or fracture propagation. These events occur within

a speciﬁc volume of rock and emit detectable seismic

waves. Figure 12(c) illustrates the key distinction between

acoustic emissions and microseismic events, which lies in

the higher seismic motion frequencies of the former com-

pared to the latter. Sensors are capable of recording the

seismic waves radiated by rock microcracks to a certain

extent. Subsequently, specialized software is employed to

interpret the data and determine the source characteristics

of microseismic events, including their time, location,

energy, magnitude, and apparent stress. Each microseismic

event contains detailed information about the stress varia-

tion within the targeted rock volume. By analyzing the

source parameters of numerous microseismic events, it

becomes possible to predict the tendency, location, and

severity of rockburst through an in-depth analysis of

microseismic data. Notably, while earthquakes and mine

rockbursts can be sensed by humans, microseismic events

associated with underground excavation can only be

detected by sensors (Ma et al., 2021).

4.5 Comments for diﬀerent forecasting methods

An important ﬁnding of the current review is that

researchers have extensively employed various intelligent

methods to address the signiﬁcant issue of rockburst pre-

diction. This trend is expected to contribute to the develop-

ment of reliable and eﬀective models in this ﬁeld. In the

context of forecasting rockburst events, approaches can

be broadly categorized into four main types: empirical,

simulation, mathematical, and microseismic monitoring

methods. Empirical models possess distinct advantages in

providing fair ﬁndings. Their ease of use for engineering

applications and straightforward form are particularly

noteworthy. However, some conventional classiﬁcation

thresholds for these models are inconsistent or even signif-

icantly diﬀerent, raising concerns about their eﬀectiveness.

Another signiﬁcant advantage of empirical criteria is their

reduced reliance on data pretreatment processes and train-

ing settings compared to mathematical or numerical meth-

ods. Simulation technology, as revealed in the review,

generally requires less data compared to mathematical

modeling techniques. It allows for easy manipulation of

inﬂuencing parameters under controlled circumstances.

Fig. 11. Suggested methods for assessing hard rock rockburst using a data-driven approach.

J. Zhou et al. / Underground Space 14 (2024) 70–98 83

However, accurately simulating defects can often be chal-

lenging due to their unpredictability and complexity, lead-

ing to signiﬁcant computational costs. Mathematical

models are evolving rapidly, with the introduction of new

and sophisticated technologies in this ﬁeld. These advance-

ments hold the potential for improvements or even signiﬁ-

cant breakthroughs in rockburst prediction. Additionally,

these models oﬀer simplicity in calculations and the ability

to forecast future scenarios. However, a notable limitation

is the requirement for a substantial amount of rockburst

data to accurately determine important model parameters.

In contrast, microseismic monitoring technologies have the

advantage of immediately and continuously collecting

essential data, providing accurate insights into the forma-

tion mechanism and evolution patterns of rockburst. This

enables timely assessment and prediction of rockburst

events. Nevertheless, deriving meaningful patterns from

large volumes of microseismic data remains a pressing chal-

lenge in this area.

Overall, comparing the superiority of diﬀerent methods

is challenging without conducting comprehensive evalua-

tions under the same conditions. Each approach is still

under development and has its own advantages and disad-

vantages. Consequently, there is no standardized method

or approach that can universally evaluate the eﬀectiveness

of all projects in underground space excavation. The level

of precision and data accessibility play crucial roles in

determining the appropriateness of a particular model for

rockburst forecasting.

5 Rockburst prevention

5.1 Rockburst support design principles and control

strategies

The reduction of rockburst risk often relies on the selec-

tion of eﬀective advance prediction methods. It is com-

monly recognized among scholars that accurate

prediction of rockburst occurrences necessitates the deter-

mination of their location, timing, intensity, and underly-

ing mechanisms. However, due to the inherent instability

of rock mass characteristics and various boundary condi-

tions (such as high in-situ stress, pressure relief, and signif-

icant disturbances), the design of rockburst support must

prioritize ground control methods rather than relying

solely on uncertain predictions. A critical line of defense

is the implementation of burst-resistant support systems,

which aim to prevent or mitigate damage and ensure oper-

ational safety. Therefore, it is crucial to thoroughly com-

prehend the principles and objectives of support design

during the initial phase of project development.

In contrast to conventional rock support methods that

focus on managing loose rock in shallow zones and pre-

venting gravity-induced rockfalls, support design in

burst-prone ground must take diﬀerent variables into

account. Dou et al. (2021) suggested that the root cause

of tunnel damage and surrounding rock deformation at

high stress levels is due to strong shock waves that cause

sharp increases in rock stress in a short period of time.

Fig. 12. Microseismic monitoring diagram. (a) Application of microseismic monitoring systems in hard rock mines, (b) microseismic monitoring ﬂow

chart, (c) seismic motion wave frequency spectrum and ﬁeld of application of acoustic emissions and microseismic events techniques, and (d) recognition

and studying of microseismic signals.

84 J. Zhou et al. / Underground Space 14 (2024) 70–98

Thus, support design in burst-prone areas must consider

the ability to withstand dynamic loads and signiﬁcant

deformations resulting from rock dilation. Engineers must

conduct a rockburst severity division, evaluate the require-

ments for rock support, and choose appropriate support

measures based on the site characteristics. Kaiser and Cai

(2012) introduced six principles of consideration for rock-

burst support, as shown in Fig. 13. The ﬁrst principle is

to avoid rockburst occurrences as much as possible by

using hefty rock support to sustain the rock without reduc-

ing its loads or counteracting its stresses. Methods based

on this principle include adjusting the size or shape of the

stope, transforming the position of the roadway, using

alternative excavation shapes, changing the mining

sequence, or switching mining techniques. The second prin-

ciple suggests that yielding support should be applied to

burst-prone ground. Brittle rock may experience signiﬁcant

impact energy and undergo signiﬁcant rock dilation when

it cracks. Therefore, the yielding support system must be

ﬂexible and able to absorb dynamic energy, as well as com-

bine with the surrounding rock as a whole to reinforce the

rock mass. The third principle emphasizes the importance

of considering the weakest link, which is typically the

retaining element. In signiﬁcant rockburst occurrences,

the contact between bolts and the mesh often fails. There-

fore, one critical element to consider is the performance of

the connections between bolts and the mesh. The choice of

surface support components and the strength of the con-

nections must be matched with the bearing capacity of

the bolts. Additionally, the design principle should incor-

porate the integrated system approach. The retaining ele-

ments and surface support elements form the

fundamental components of a rock support system. How-

ever, some support components may only function eﬀec-

tively in speciﬁc environments. Therefore, it is necessary

to integrate all support components into a cohesive and

integrated system. The ﬁfth principle emphasizes simplicity

and cost control in design. Clients are unlikely to adopt

rock support products that are complex and expensive to

manufacture. Simplicity also contributes to eﬃciency.

While many businesses strive to reduce costs and stream-

line operations to enhance competitiveness, they cannot

compromise on safety. Statistics indicate that the ﬁnancial

losses resulting from security incidents far outweigh the

beneﬁts of cost control. In other words, the cost of prevent-

ing burst-prone grounds is signiﬁcantly lower than the eco-

nomic losses caused by rockburst accidents. The ﬁnal

principle is the predictability and adaptability of the rock

support system. It is unrealistic to create a design that can-

not be modiﬁed, as the burst-prone ground environment

and the classiﬁcation of rockburst severity are constantly

changing. Therefore, a successful strategy for managing

burst-prone ground involves a well-adapted and ﬂexible

rock support system. These fundamental principles serve

as a guide for steering support design in the right direction

for speciﬁc projects.

According to Kaiser and Cai (2012), the comprehensive

approach employed in managing the entire mining or tun-

neling process is referred to as the strategy in underground

excavation. In essence, the utilization of diverse tools or

methods to address ﬁeld problems necessitates strategic

thinking. Rockburst control strategies can be classiﬁed into

two categories: active induction and passive protection.

The passive support method comprises three key compo-

nents: ground preconditioning, alternate mining tech-

niques, and rockburst support. Active induction aims to

harness the high stress and stored energy within and sur-

rounding deep hard rock, employing regulation and con-

trolled release to enhance the rock environment and

induce controlled self-fracturing. The overall control strat-

egy is illustrated in Fig. 14.

To address dynamic instability issues, the core element is

the risk management strategy, which is an extension of the

control strategy. Notably, Diederichs (2018) introduced the

well-known ‘‘Butterﬂy”risk management strategy for rock-

burst in hard rock excavations (refer to Fig. 15). The but-

terﬂy risk management consists of two components: the

propensity to disaster and its resulting consequences. The

hazard potential component identiﬁes rockburst triggers

and outlines a contractor control sequence from top to bot-

tom. The consequence aspect addresses potential hazards

and prioritizes their handling once they occur. The imple-

mentation of a microseismic monitoring system is crucial

for situational awareness in advance. Additionally, incor-

porating ‘‘Management”measures is recommended to mit-

igate the impact of dynamic fractures.

A crucial aspect of disaster prevention and control in

burst-prone zones involves the timely implementation of

rockburst prevention and control measures. Failure to do

so can result in a signiﬁcant increase in support costs. Dur-

ing the design phase of underground excavation, it is essen-

tial to accurately estimate the occurrence time of rockburst

to enhance the eﬀectiveness and practicality of prevention

and control strategies. While various techniques for rock-

burst prevention and control are available, further investi-

gation is needed to determine the eﬃcacy of speciﬁc

measures. Therefore, in order to eﬀectively guide support

work in burst-prone areas, it is necessary to outline the

speciﬁc actions and contents of rockburst risk prevention

and management. Figure 16 presents a comprehensive

guide for preventing and controlling rockburst in deep

hard rock mines. This ﬁgure illustrates the ﬁve processes

involved in rockburst risk prevention and control, namely

type identiﬁcation, risk assessment, selection of prevention

and control methods, design work, and eﬃcacy veriﬁca-

tion. The speciﬁc details of each process are represented

by the corresponding color blocks below the ﬁgure.

5.2 Rockburst support methods

It is well-established that mining-induced rockburst is a

complex phenomenon that frequently occurs during deep

J. Zhou et al. / Underground Space 14 (2024) 70–98 85

underground construction. Extensive eﬀorts have been

devoted to understanding the causes of rockburst and

devising eﬀective support strategies for burst-prone zones.

Consequently, a substantial body of research has been con-

ducted on rockburst support methods. The widely utilized

‘‘New Austrian Method”is commonly adopted for tunnel

support design (Kolymbas, 2005; Meng et al., 2011;

Rehman et al., 2020). Independently, Pan et al. (2021)

developed the asymmetric support method of full-section

anchor cable spraying, which signiﬁcantly enhances the

ability of the surrounding rock in roadways to withstand

rockburst damage and further elucidates the principle of

asymmetric damage. Wu et al. (2021) introduced a new

synergistic support method for deep hard rocks, employing

the synergistic approach of ‘‘pressure relief - support – pro-

tection”to eﬀectively enhance the rock bolting capacity. Li

et al. (2007) examined the damage characteristics of the

surrounding rock in a stope roadway following a

rockburst, providing insights into how rockbolt support

maintains stress balance and supports the roadway. Subse-

quently, a high-strength composite support method was

designed.

Currently, rockbolts are extensively employed to rein-

force underground excavation projects, making them the

primary safety protection method. Conventional bolts are

typically characterized by their load-bearing capacity or

ductility. However, due to the dynamic and destructive nat-

ure of rockburst events, it is essential for the supporting

structure to possess robustness to withstand the kinetic

energy of ejected rock and toughness to absorb energy

emitted from the rock mass. Evidently, the conventional

method is not the ideal choice for rockburst support due

to its limited energy absorption capacity. In contrast,

energy-absorbing rockbolts have emerged as a novel type

of support device that has garnered considerable attention.

These bolts exhibit a high energy-absorbing capacity,

Fig. 13. Six basic rockburst support design principles based on ﬁeld experience.

86 J. Zhou et al. / Underground Space 14 (2024) 70–98

enabling them to bear substantial loads and accommodate

signiﬁcant rock displacement. Consequently, they prove to

be eﬀective support mechanisms during dynamic failure

processes.

Energy-absorbing bolts oﬀer a dual advantage of high

deformation capacity and resistance. On one hand, they

can absorb a signiﬁcant amount of energy from the sur-

rounding rock before reaching equilibrium. On the other

hand, they reinforce the rock mass and enhance its

mechanical condition. The development of energy-

absorbing bolts has attracted considerable attention from

scholars (Wang et al., 2020). Bolts with strong energy

absorption capacity have gained international recognition

as essential components of underground engineering sup-

port systems (Wu et al., 2022). Since their invention,

researchers have continuously optimized and improved

each component of energy-absorbing bolts to enhance their

eﬀectiveness. Figure 17 illustrates some typical energy-

absorbing bolt designs. These bolts dissipate energy

through buﬀering interactions with the surrounding rock,

eﬀectively reducing impact energy and forming a cohesive

unit with the rock mass to withstand impact loads. This

Fig. 14. Rockburst control strategy.

Fig. 15. Management of risk of rockburst in underground space excavations. (Modiﬁed from Diederichs, 2018).

J. Zhou et al. / Underground Space 14 (2024) 70–98 87

ensures the safety of underground construction projects

and maintains the stability of the support system. Further-

more, signiﬁcant advancements have been made in the

structural design of rockbolts. Jing et al. (2020) investi-

gated the load-bearing changes and evolutionary character-

istics of deformation damage in tunnel surrounding rock

Fig. 16. Guidance on rockburst risk prevention and control in underground space excavation. Blue indicates rockburst type identiﬁcation; green indicates

rockburst risk assessment; yellow indicates prevention and control measure selection; pink indicates prevention and control measure design; purple

indicates prevention and control eﬀectiveness veriﬁcation.

88 J. Zhou et al. / Underground Space 14 (2024) 70–98

after the installation of rockbolts, obtaining process curves

for structural deformation and fracture damage. Jiao and

Ju (2021), based on the strength characteristics, impact

damage, and stress state changes of rock bolt weighing

members, derived an impact damage criterion for bolts.

Huang et al. (2020) studied the deformation mode and pro-

posed a mechanism for deep large deformation damage in

anchored structures. Wang et al. (2014) examined the tan-

gential stress acting on the surrounding rock and investi-

gated the overall structural mechanics law of rock bolts

and the support eﬀect on the surrounding rock as a whole.

Looking ahead, future underground construction projects

will involve even deeper excavations and challenging envi-

ronments, characterized by high stress, high temperature,

and great lifting depth. These conditions will test the capa-

bilities of support technologies.

5.3 Safe and eﬃcient excavation concept for deep hard rock

projects

Deep hard rock environments are characterized by high

levels of stress, and the process of deep rock excavation

involves rapid unloading of the highly stressed rock mass,

accompanied by dynamic disturbances such as blasting

impact, large-scale collapses, and even seismic activity in

the surrounding mining area. The key to technological

advancement in deep hard rock mining lies in transforming

the unconventional rock fractures induced by mining dis-

turbances in deep, high-stress rock into controlled frac-

tures, and eﬀectively regulating the energy release of the

deep, high-stress rock. Li et al., (2017b) proposed the con-

cept of ‘‘turning harm into proﬁt”for the safe and eﬃcient

exploitation of deep earth resources, as depicted in Fig. 18.

The three primary factors inﬂuencing deep mining and dis-

aster mitigation are high stress, high ground temperature,

and great lifting depth. These factors can be transformed

from sources of catastrophe into advantageous factors.

High stress facilitates the fracturing of hard rock and

enables better control of rock blocks. High ground temper-

ature accelerates the interaction between minerals and

leaching ﬂuids in in-situ leach mining. The high water pres-

sure in great lifting depth promotes the use of high-pressure

equipment or underground power sources for eﬃcient

water supply. Figure 19 presents an overview of the strate-

gies to achieve an eﬀective transformation of these three

challenges.

6 Future directions

In the foreseeable future, the ﬁeld of underground exca-

vation is expected to undergo a rapid transition towards

intelligent technologies. Figure 20 illustrates various

emerging technologies such as deep learning, intelligent

robots, digital technology, 5G networks + VR/AR, Inter-

net of Things, and digital twins, which will progressively

play a crucial role in rockburst prediction and prevention.

The integration of these technologies will enhance the

precision and eﬀectiveness of rockburst prediction and pre-

vention, reduce the workload and risks faced by miners,

and foster the sustainable development of the mining

industry (Li et al., 2018; Zhang et al., 2020a; Tang et al.,

2019; Pan & Zhang, 2021).

(1) Deep learning

Deep learning (DL), a cutting-edge data-driven technol-

ogy, has recently shown promise in engineering applica-

tions, presenting both opportunities and challenges.

Compared to traditional methods, DL algorithms have

demonstrated remarkable predictive capabilities across

various ﬁelds and hold great potential for rockburst predic-

tion (Bergen et al., 2019; Merghadi et al., 2020;Yu & Ma,

2021). Mining data typically encompass geological infor-

mation, groundwater conditions, ground stress, blasting

parameters, and more, which can be analyzed and modeled

using DL algorithms to enhance the accuracy of rockburst

prediction. For instance, DL models can be employed to

classify and cluster geological data, identifying potential

anomalies in the structural behavior of rock masses and

enabling the forecast of rockburst probability and severity.

Furthermore, DL algorithms can automatically segment

and identify rock masses, thereby improving the accuracy

and eﬃciency of mine design.

(2) Intelligent robots

Intelligent robots are experiencing rapid advancements

and are increasingly being adopted in hard rock engineer-

ing for semi-automatic or fully automated applications.

Two primary types of robots are utilized: ground-based

robots and aerial robots. The widespread use of intelligent

robots has brought signiﬁcant beneﬁts to mine inspection,

mining operations, and blasting activities. These robots

aid in better predicting and avoiding rockburst incidents,

while facilitating the identiﬁcation of potential rock-

related issues. Equipped with sensors and cameras, intelli-

gent robots can continuously monitor the conditions of

the rock mass and the underground environment. The

gathered information is then relayed to miners for analysis

and decision-making processes. Furthermore, the autono-

mous inspections and processing capabilities of intelligent

robots alleviate the workload and risks faced by miners.

(3) Digital technology

The increasing convergence and evolution of new digital

technologies and applications have resulted in the frequent

utilization of digital creation, digital transactions, and dig-

ital assets. Empowered by digital technology, traditional

industries are swiftly undergoing digital transformation,

bringing forth new revolutions across various sectors. In

the ﬁeld of mining engineering, digital technology has

gained widespread adoption and provides crucial data

and information to support rockburst prediction and

J. Zhou et al. / Underground Space 14 (2024) 70–98 89

prevention. For instance, digital technology enables the

creation of 3D geological models, allowing for accurate

analysis and prediction of rock distribution and structure.

Additionally, digital technology facilitates the optimization

and adjustment of blasting parameters, thereby reducing

the risk of rockburst incidents.

(4) 5G networks + VR/AR

Leveraging the advancements in ﬁfth-generation net-

work technology, the combination of Virtual Reality

(VR) and Augmented Reality (AR) is rapidly evolving,

oﬀering enhanced user experiences. The integration of 5G

networks with VR/AR technologies has found extensive

application in mine design and training, enabling miners

to better comprehend the rock mass conditions and mine

environment, thereby improving their ability to prevent

rockburst incidents. For instance, VR technology can

establish a comprehensive mine scene model and provide

immersive virtual reality displays, allowing miners to intu-

itively perceive the rock structure and underground envi-

ronment. Moreover, AR technology facilitates intelligent

management and monitoring of mine equipment and tools,

thereby reducing the risk of rockburst occurrences.

(5) Internet of Things (IoT) technology

IoT refers to a network of interconnected physical

devices that collect real-time operational data and

resources within a project. The utilization of IoT technol-

ogy in mine monitoring and management has become a

common practice, as it enables equipment connectivity,

Fig. 17. History of energy absorbing bolts.

Fig. 18. Framework for the development of safe and eﬃcient exploitation from deep earth resources.

90 J. Zhou et al. / Underground Space 14 (2024) 70–98

real-time monitoring, and enhances the capacity for rock-

burst prediction and prevention. For instance, IoT technol-

ogy enables the networking and data sharing among

sensors and monitoring equipment within mines, enabling

real-time monitoring and early warning systems for the

underground environment and the state of the rock mass.

Additionally, IoT technology facilitates the tracking and

localization of miners and equipment, timely identiﬁcation

and treatment of engineering disasters, and improves the

eﬃciency and accuracy of rockburst prediction and

prevention.

(6) Digital Twins

Digital twin technology, which digitizes the physical

world and creates a three-dimensional model, enables

real-time monitoring and simulation (Pan & Zhang, 2021;

Min et al., 2019). By employing digital twin technology,

miners can gain a comprehensive understanding of the sub-

surface environment and the condition of the rock mass,

thereby enhancing their capabilities for rockburst predic-

tion and prevention. Speciﬁcally, digital twin technology

can establish a virtual representation of the mine, enabling

real-time monitoring and simulation of the state and

changes within the rock mass, facilitating the estimation

of rockburst possibilities. Additionally, digital twin tech-

nology enables the simulation and optimization of blasting

parameters and engineering schemes, thereby enhancing

the accuracy and eﬃciency of rockburst prediction and

prevention.

7 Conclusion

A growing body of research has emerged in recent years

to address the global challenge of rockburst risk reduction.

To comprehensively understand the current state and pro-

gress of this ﬁeld, it is imperative to conduct thorough and

quantitative literature studies. One commonly used tool for

analyzing global trends since 1990 is the software CiteS-

pace. Analysis of publications reveals that China, the US,

Australia, Canada, and Russia are the top ﬁve countries

contributing signiﬁcantly to the ﬁeld. This indicates their

notable contributions to rockburst research. Furthermore,

an examination of co-institutions demonstrates that while

global communication has strengthened with the rapid

expansion of the Internet, institutional cooperation primar-

ily occurs among organizations and disciplines with similar

interests, with limited interdisciplinary and cross-

disciplinary collaboration. Notably, the analysis of the

timeline graph indicates that prediction is currently the

most prominent research area, with a strong connection

to artiﬁcial intelligence.

In rockburst classiﬁcation, selecting the appropriate cri-

teria is crucial due to the existence of numerous indicators.

Diﬀerent criteria are applied depending on factors such as

lithology, geological zone, and burial depth. Therefore, it is

essential to systematically document the basic information

of typical rockburst incidents and statistically analyze their

characteristics. This will provide a comprehensive

understanding of the mechanisms and conditions that

trigger rockburst events and facilitate the improvement of

Fig. 19. ‘‘Turning harm into proﬁt”strategy for the three high environments.

J. Zhou et al. / Underground Space 14 (2024) 70–98 91

classiﬁcation criteria. Based on this foundation, it can be

inferred that hard rocks, including granite, gneiss, horn-

blende, silica, and other hard brittle rocks, are typically

found at depths of 300 meters or more where rockburst

phenomena occur.

As the depth of excavations increases, the occurrence of

dynamic rock failure events, such as rockburst, becomes

more likely. Extensive research has been conducted to

investigate the causes and mechanisms of rockburst. One

triggering mechanism of rockburst is the increase in tan-

gential stress and decrease in radial stress caused by exca-

vation. Various approaches for predicting rockburst in

underground engineering have been explored, including

empirical criteria, numerical simulations, mathematical

methods, microseismic monitoring, and rockburst charts.

A comprehensive review of these approaches is presented.

However, comparing the superiority of these methods

under the same conditions is challenging as each method

is still in development and has its own advantages and lim-

itations. Furthermore, as excavations reach greater depths,

severe rockburst events become inevitable and preventive

strategies must be regularly employed. Understanding the

mechanics of rockburst and identifying the key parameters

that aﬀect rockburst damage are crucial steps in designing

eﬀective rockburst support systems. Additionally, it is

essential to comprehend two key prevention strategies,

namely rockburst control strategy and butterﬂy risk man-

agement approach, along with the seven design principles

for rockburst support. It is of utmost importance that

any rockburst support design adheres to guidelines for

preventing rockburst risk. By outlining concrete actions

and considerations for rockburst risk prevention and

management, the practical aspects of support design in

burst-prone grounds can be deﬁned. Lastly, acknowledging

Fig. 20. Directions for the development of underground space excavation based on intelligent technology.

92 J. Zhou et al. / Underground Space 14 (2024) 70–98

the negative eﬀects of subterranean excavation, including

high ground temperature, great lifting depth, and high

stress, the design process for rock support should aim to

transform the challenging excavation environment into

favorable and controllable elements. However, due to the

unpredictable nature of dynamic destruction and the

uncertain surrounding contributing factors, there is still

limited conﬁdence in predictive means and preventive

measures.

In recent decades, digital technology has rapidly

advanced, and the utilization of big data has expanded in

the ﬁeld of traditional underground engineering. Particu-

larly, the application of cutting-edge technologies has gar-

nered signiﬁcant attention. It has been observed that

various cutting-edge methodologies have made substantial

contributions to the modernization of the underground

engineering industry, resulting in a more reliable, auto-

mated, time-saving, and cost-eﬀective excavation process.

Implementing future directions that incorporate cutting-

edge science, such as DL, IoT, sensing technology, and

more, can decrease the risk of engineering disasters. These

directions can assist project decision-makers in eﬀectively

managing the challenging and hazardous tasks associated

with excavation while promoting the development of intel-

ligent disaster mitigation and preparedness.

Declaration of competing interest

The authors declare that they have no known competing

ﬁnancial interests or personal relationships that could have

appeared to inﬂuence the work reported in this paper.

Acknowledgement

This research is partially supported by the National

Natural Science Foundation Project of China (Grant

Nos. 42177164 and 72088101), and the Distinguished

Youth Science Foundation of Hunan Province of China

(Grant No. 2022JJ10073).

Appendix A Statistics on rockburst examples of typical hard rock projects in China

Projects Burial

depth (m)

Rock properties Rockburst phenomenon

Tongguan Gold Mine,

Shaanxi, China

360–730 Mixed rocks, mixed granites, plagioclase,

gneisses, f = 10–16, complex tectonic

stresses, brittle and hard.

Minor rockburst, where rock fragments

are dislodged, often with loud noises and

local ejections from the surrounding rock.

Yangtze Valley Mine,

Henan Qinling

Mining Company,

China

335–450 Quartz-mica gneiss, medium to coarse

grained structure, gneissic. The lithology is

hard and brittle with well-developed

joints.

After compression and expansion of the

450 m level, a sudden rock burst occurred

on the roof, resulting in a large area of

ﬂaky gang within 10 m of the shaft above

and a large area of sprayed protective

anchor mesh falling oﬀ..

Baima Iron Ore Mine,

China

400–600 Rocks of high strength and good integrity,

hornblende orthoclase.

Rockburst occurs from the construction

surface 5 to 8 m, 8 m away from the

strongest performance. The form of

expression is shear damage and splitting

damage 2 kinds, mainly ejected and burst

2 kinds of phenomena.

Chengchao Iron Ore

Mine, China

430–568 Mainly amphibolite, amphibolite, dacite,

diorite, granite, granite porphyry and

quartz amphibolite.

Rockburst occurs in shafts or chambers in

areas of concentrated stress due to large

scale blasting vibrations during shaft

development and quarry cutting work.

Hongtushan Copper

Mine, China

400–1077 Mixed granites, gneisses, metamorphic

rocks, high tectonic stresses.

The main forms of rockburst are pit

ﬂakes, block ejections and roof falls. A

strong rockburst occurs when a sound

similar to a large blast is emitted and the

ground is usually felt as a distinct tremor.

(continued on next page)

J. Zhou et al. / Underground Space 14 (2024) 70–98 93

Appendix A (continued)

Projects Burial

depth (m)

Rock properties Rockburst phenomenon

Selenium Flag River

Phosphate Mine,

China

518 Granite greenstone. The intensity and frequency of rockburst

increased signiﬁcantly as the project

progressed, with varying degrees of

rockburst occurring during the

construction of the 518-deep ramp below

the shaft, and creating a great potential for

construction safety.

Waya IV Mine, China 586.18 This is a brittle, hard, siliceous dolomite.

The ﬁssures are mostly ﬁlled with calcite

and other impurities due to the inﬂuence

of faults, and the joints are well developed.

Rockburst accompany the entire

production process, and are frequent

during the process of back mining and

roadway boring. Rockburst and slagging

and roof eruptions occur from time to

time, accompanied by a gradual increase

in the size of the mining area.

Sanshan Island Gold

Mine, China

555–960 Granite. Rockburst such as rock rattles and small

rock ejections occurred during the

construction of the southern ﬂank of the

555 m middle section and the 960 m

middle section ramp at the mine.

Ashel Copper Mine,

China

550–1100 Elastic energy index Wet = 2.3, pyrite

(88.74 MPa), brittleness factor B

= 25.95.

The manifestations are cracking of

concrete spraying in the roadway, collapse

of large pieces of the surrounding rock,

shelling of the surrounding rock,

deformation of the supporting steel arch

and brittle ejection of the surrounding

rock.

Chener Gold Mine,

China

580–943 Quartz veins It occurred in a large section, causing

rapid ejection of broken rock, ﬂaking oﬀ

on both sides of the roadway, and severe

dislodgement of the top of the constructed

roadway.

Xinjiang Water

Diversion Tunnel,

China

600–720 Metamorphic granite, amphibolite and

mica-quartz schist are medium to hard

rocks of average strength, with

undeveloped joints and ﬁssures and intact

rock masses.

Rockburst begins along the weak points of

the section joints, the rock falls in response

to the sound, followed by the sound of

ﬂake and block spalling.

Bayu Tunnel, China 3260–

5500

Hornblende Black Cloud Granite. The main manifestations are the bending

and bulging of rock slabs caused by tensile

action and the spalling of rock ﬂakes

caused by shear action.

Shuangjiangkou

Hydropower

Station, China

2300 Granite pegmatite veins. The rocks have a

mean uniaxial compressive strength of

139 MPa.

The roof arch of the branch cavern has

suﬀered a more serious rockburst.

Cao Guo Shan Tunnel,

China

903.8 Granite. At K35 + 480 (burial depth 507 m) there

was a slight block drop and abnormal

sound, which is a slight rockburst

phenomenon.

94 J. Zhou et al. / Underground Space 14 (2024) 70–98

Appendix A (continued)

Projects Burial

depth (m)

Rock properties Rockburst phenomenon

Qinling Water

Transfer Tunnel,

China

1260 Semipelagic rocks. Average ﬁeld test rock

strengths range from 75 to 230 MPa.

Four strong rockbursts occurred

consecutively on the excavation face, with

the bursting of the slag body mostly in

large pieces

Jinping Secondary

Hydropower

Station, China

1500–

2000

Marl, limestone, sandstone and slate.

Rocks are relatively intact. Uniaxial

compressive strength of the rock ranges

from 55 to 114 MPa.

A number of rockbursts occurred during

the construction of the tunnel, mainly in

the arch of the cavern, and most of the

rock blocks were spalling in ﬂakes and

layers, with rough fracture surfaces,

showing tensional damage.

Sanzuling Tunnel,

Lalin Railway,

China

450 Granite and amphibolite. There are local

stress concentrations after excavation.

The presence of rock crumbling

phenomenon, the area is generally in the

2–4 m

, individual more than 8 m

, burst

pit depth of about 50 cm, mostly occurring

near the arch waist.

Gangmu La Mountain

Tunnel, China

300–760 Clinopyroxene, amphibolite and gneiss. This is manifested by the loosening and

stripping of the structural face, forming a

wedge or trapezoidal shaped burst pit in

the wall of the surrounding rock.

References

Adoko, A. C., Gokceoglu, C., Wu, L., & Zuo, Q. J. (2013). Knowledge-

based and data-driven fuzzy modeling for rockburst prediction.

International Journal of Rock Mechanics and Mining Sciences, 61,

86–95.

Afraei, S., Shahriar, K., & Madani, S. H. (2019). Developing intelligent

classiﬁcation models for rock burst prediction after recognizing

signiﬁcant predictor variables, section 2: Designing classiﬁers. Tun-

nelling and Underground Space Technology, 84, 522–537.

Askaripour, M., Saeidi, A., Rouleau, A., & Mercier-Langevin, P. (2022).

Rockburst in underground excavations: A review of mechanism,

classiﬁcation, and prediction methods. Underground Space, 7(4),

577–607.

Bardet, J. P. (1989). Finite element analysis of rockburst as surface

instability. Computers and Geotechnics, 8(3), 177–193.

Barton, N., Lien, R., & Lunde, J. (1974). Engineering classiﬁcation of rock

masses for the design of tunnel support. Rock Mechanics, 6(4),

189–236.

Bergen, K. J., Johnson, P. A., de Hoop, M. V., & Beroza, G. C. (2019).

Machine learning for data-driven discovery in solid Earth geoscience.

Science, 363(6433), eaau0323.

Blake, W., & Hedley, D. G. F. (2003). Rockbursts: Case studies from

North American hard-rock mines. Littleton, CO: Society for Mining,

Metallurgy, and Exploration. Inc.

Brown, E. T., & Hoek, E. (1980). Underground excavations in rock. CRC

Press.

Cai, M., Kaiser, P. K., Morioka, H., Minami, M., Maejima, T., Tasaka,

Y., & Kurose, H. (2007). FLAC/PFC coupled numerical simulation of

AE in large-scale underground excavations. International Journal of

Rock Mechanics and Mining Sciences, 44(4), 550–564.

Cai, M. (2013). Principles of rock support in burst-prone ground.

Tunnelling & Underground Space Technology Incorporating Trenchless

Technology Research, 36, 46–56.

Chen, B. R., Feng, X. T., Li, Q. P., Luo, R. Z., & Li, S. (2015). Rock burst

intensity classiﬁcation based on the radiated energy with damage

intensity at Jinping II hydropower station, China. Rock Mechanics and

Rock Engineering, 48, 289–303.

Chen, C. (2006). CiteSpace II: Detecting and visualizing emerging trends

and transient patterns in scientiﬁc literature. Journal of the American

Society for information Science and Technology, 57(3), 359–377.

Chen, C., & Zhou, J. (2023). A new empirical chart for coal burst liability

classiﬁcation using Kriging method. Journal of Central South Univer-

sity, 30(4), 1205–1216.

Chen, H. J., Li, X. B., & Zhang, Y. (2008). Study on application of set pair

analysis method to prediction of rockburst. Journal of University of

South China, 22(4), 10–14 (in Chinese).

Cook, N. G. W. (1983). Origin of rockbursts. In Rockbursts: prediction

and control. Symposium (pp. 1–9).

Cook, N. G. W. (1965). A note on rockbursts considered as a problem of

stability. Journal of the Southern African Institute of Mining and

Metallurgy, 65(8), 437–446.

Cook, N. G. W. (1963). The basic mechanics of rockbursts. Journal of the

Southern African Institute of Mining and Metallurgy, 64(3), 71–81.

Deng, J., & Gu, D. S. (2018). Buckling mechanism of pillar rockbursts in

underground hard rock mining. Geomechanics and Geoengineering, 13

(3), 168–183.

Diederichs, M. S. (2018). Early assessment of dynamic rupture hazard for

rockburst risk management in deep tunnel projects. Journal of the

Southern African Institute of Mining and Metallurgy, 118(3), 193–204.

Dou, L., Chen, T., Gong, S., He, H., & Zhang, S. (2012). Rockburst

hazard determination by using computed tomography technology in

deep workface. Safety Science, 50(4), 736–740.

Dou, L., Zhou, K., Song, S., Cao, A., Cui, H., Gong, S., & Ma, X. (2021).

Occurrence mechanism, monitoring and prevention technology of

rockburst in coal mines. Journal of Engineering Geology, 29(4),

917–932.

Du, K., Li, X., Su, R., Tao, M., Lv, S., Luo, J., & Zhou, J. (2022). Shape

ratio eﬀects on the mechanical characteristics of rectangular prism

rocks and isolated pillars under uniaxial compression. International

Journal of Mining Science and Technology, 32(2), 347–362.

Du, K., Tao, M., Li, X. B., & Zhou, J. (2016). Experimental study of

slabbing and rockburst induced by true-triaxial unloading and local

dynamic disturbance. Rock Mechanics and Rock Engineering, 49(9),

3437–3453.

Feng, X. T., Chen, S., & Zhou, H. (2004). Real-time computerized

tomography (CT) experiments on sandstone damage evolution during

triaxial compression with chemical corrosion. International Journal of

Rock Mechanics and Mining Sciences, 41(2), 181–192.

Feng, X. T., Xue, Y. S., & Feng, G. L. (2012). Mechanism, warning and

dynamic control of rockburst evolution process pp. ISRM-ARMS7.

ISRM International Symposium-Asian Rock Mechanics Symposium.

ISRM.

J. Zhou et al. / Underground Space 14 (2024) 70–98 95

Feng, X. T., Chen, B. R., Zhang, C. Q., Li, S. J., & Wu, S. Y. (2013).

Mechanism, Warning and Dynamical Control of Rockburst Evolution

Process. Beijing: Science Press (pp. 380–391). Beijing: Science Press (in

Chinese).

Gu, R. (2013). Distinct element model analyses of unstable failures in

underground coal mines. Colorado School of Mines.

Guo, C., Zhang, Y. S., Deng, H., Su, Z., & Sun, D. (2011). Study on rock

burst prediction in the deep-buried tunnel at Gaoligong Mountain

based on the rock proneness. Geotech Invest Survey, 39(10), 8–13, in

Chinese).

He, S., Song, D., Mitri, H., He, X., Chen, J., Li, Z., Xue, Y., & Chen, T.

(2021). Integrated rockburst early warning model based on fuzzy

comprehensive evaluation method. International Journal of Rock

Mechanics and Mining Sciences, 142, 104767.

Heal, D. (2010). Observations and analysis of incidences of rockburst

damage in underground mines [Ph. D. Thesis]. Australia: University of

Western Australia.

Heal, D., Hudyma, M., & Potvin, Y. (2006). Evaluating rockburst damage

potential in underground mining. Golden Rocks 2006, The 41st US

Symposium on Rock Mechanics (USRMS). OnePetro.

Hou, F. L., & Wang, M. Q. (1989). Criterion and prevention measures on

rockburst in circular tunnel. Proceedings of the 2nd national rock

mechanics and engineering (Chinese Journal of Rock Mechanics and

Engineering). Beijing: Knowledge Press (in Chinese).

Huang, B. X., Zhang, N., & Jing, H. W. (2020). Large deformation theory

of surrounding rock rheology and structural instability of deep mining

roadway. Chinese Journal of Coal, 45(03), 911–926 (in Chinese).

Hudson, J. A., & Feng, X. T. (2015). Rock engineering risk (Vol. 1). CRC

Press.

Ishida Shimizu Murata & Kanagawa (2009). Importance of inhomogene-

ity in rock fracturing deduced from distinct element simulation and in-

situ direct shear test. In Proceedings of the 7th International Symposium

on Rockburst and Seismicity in Mines (Keynote Presentation), Dalian,

China (pp. 3–18).

Jia, Y. P., Lv, Q., Shang, Y. Q., Du, L. L., & Zhi, M. M. (2014).

Rockburst prediction based on rough set and ideal point method.

Journal of Zhejiang University (Engineering Science), 48(3), 498–503

(in Chinese).

Jiang, L., Kong, P., Zhang, P., Shu, J., Wang, Q., Chen, L., & Wu, Q.

(2020). Dynamic analysis of the rock burst potential of a longwall

panel intersecting with a fault. Rock Mechanics and Rock Engineering,

53, 1737–1754.

Jiang, Q., Feng, X. T., Xiang, T. B., & Su, G. S. (2010). Rockburst

characteristics and numerical simulation based on a new energy index:

A case study of a tunnel at 2,500 m depth. Bulletin of engineering

geology and the environment, 69, 381–388.

Jiao, J. K., & Ju, W. J. (2021). Impact failure mechanism of roadway

anchorage bearing structure under dynamic load disturbance. Journal

of China Coal Society, 46, 94–105 (in Chinese).

Jing, H. W., Yin, Q., Zhu, D., Sun, Y. J., & Wang, B. (2020).

Experimental study on the whole process of instability and failure of

anchorage structure in surrounding rock of deep-buried roadway.

Journal of China Coal Society, 45(3), 889–901 (in Chinese).

Jin, Y. N., Li, X. B., Liu, P. J., & Guo, Y. (2017). Predictive liability of the

rockburst classiﬁcation based on the improved unascertained cluster-

ing model. Journal of Safety and Environment, 17(1), 12–16.

Kaiser, P. K., & Cai, M. (2012). Design of rock support system under

rockburst condition. Journal of Rock Mechanics and Geotechnical

Engineering, 4(3), 215–227.

Kaiser, P. K., McCreath, D., & Tannant, D. (1996). Canadian Rockburst

Support Handbook. Geomechanics Research Centre: Laurentian

University.

Kaiser, P. K., Tannant, D. D., McCreath, D. R., & Jesenak, P. (1992).

Rockburst damage assessment procedure. In International symposium

on rock support (pp. 639–647).

Keneti, A., & Sainsbury, B. A. (2018). Review of published rockburst

events and their contributing factors. Engineering Geology, 246,

361–373.

Ke, W. (2021). Overview of State-of-Art of Rockburst Prediction and

Prevention Techniques for Deep-buried Tunnels. Tunnel Construction,

41(02), 212–224.

Kidybin

´ski, A. (1981). Bursting liability indices of coal. In International

Journal of Rock Mechanics and Mining Sciences & Geomechanics

Abstracts (Vol. 18, No. 4, pp. 295–304). Pergamon.

Kolymbas, D. (2005). The new Austrian tunnelling method. Tunnelling

and Tunnel Mechanics: A Rational Approach to Tunnelling, 171–175.

Leger, J. P. (1991). Trends and causes of fatalities in South African mines.

Safety science, 14(3–4), 169–185.

Li, C., Zhou, J., Armaghani, D. J., & Li, X. (2021). Stability analysis of

underground mine hard rock pillars via combination of ﬁnite diﬀerence

methods, neural networks, and Monte Carlo simulation techniques.

Underground Space, 6(4), 379–395.

Li, D., Liu, Z., Armaghani, D. J., Xiao, P., & Zhou, J. (2022a). Novel

ensemble tree solution for rockburst prediction using deep forest.

Mathematics, 10(5), 787.

Li, D., Liu, Z., Xiao, P., Zhou, J., & Armaghani, D. J. (2022b). Intelligent

rockburst prediction model with sample category balance using

feedforward neural network and Bayesian optimization. Underground

Space, 7(5), 833–846.

Li, D., Liu, Z., Armaghani, D. J., Xiao, P., & Zhou, J. (2022c). Novel

ensemble intelligence methodologies for rockburst assessment in

complex and variable environments. Scientiﬁc Reports, 12(1), 1–23.

Li, G. F., Cai, J., & Guo, Z. B. (2007). Research and application of bolt

and grouting combined support technology to deep mine soft rock

roadway. Coal Science and Technology, 35(4), 44–46 (in Chinese).

Li, L. G., Wensheng, X., Yingnian, X., & Yuanhan, W. (2001).

Experimental study on simulation materials of rockburst. Journal-

Huazhong University of Science and Technology Chinese Edition, 29(6),

80–82.

Li, N., Feng, X., & Jimenez, R. (2017a). Predicting rock burst hazard with

incomplete data using Bayesian networks. Tunnelling and Underground

Space Technology, 61, 61–70.

Li, N., & Jimenez, R. (2018). A logistic regression classiﬁer for long-term

probabilistic prediction of rock burst hazard. Natural Hazards, 90(1),

197–215.

Li, X., Yi, W., Chi, H. L., Wang, X., & Chan, A. P. (2018). A critical

review of virtual and augmented reality (VR/AR) applications in

construction safety. Automation in Construction, 86, 150–162.

Li, X. B., & Dong, L. J. (2014). An eﬃcient closed-form solution for

acoustic emission source location in three-dimensional structures. AIP

Advances, 4(2), 027110.

Li, X. B., Gong, F. Q., Wang, S. F., Li, D. Y., Tao, M., Zhou, J., Huang,

L. Q., Ma, C. D., Du, K., & Feng, F. (2019). Coupled static-dynamic

loading mechanical mechanism and dynamic criterion of rockburst in

deep hard rock mines. Chinese Journal of Rock Mechanics and

Engineering, 38(4), 708–723 (in Chinese).

Li, X. B., & Gu, D. S. (2002). Disaster control and fracture mutagenesis of

high stress in mining of deep hard deposits. Paper presented at the

175th Xiangshan Scientiﬁc Conference: Frontiers and the Future of

Science: Volume 6. Beijing: China Environmental Science Press.

Li, X. B., Zhou, J., Wang, S. F., & Liu, B. (2017b). Review and practice of

deep mining for solid mineral resources. The Chinese Journal of

Nonferrous Metals, 27(6), 1236 (in Chinese).

Liang, W., Zhao, G., Wu, H., & Dai, B. (2019). Risk assessment of

rockburst via an extended MABAC method under fuzzy environment.

Tunnelling and Underground Space Technology, 83, 533–544.

Liu, J., Liu, C., Yao, Q., & Si, G. (2020). The position of hydraulic

fracturing to initiate vertical fractures in hard hanging roof for stress

relief. International Journal of Rock Mechanics and Mining Sciences,

132, 104328.

Liu, R., Ye, Y., Hu, N., Chen, H., & Wang, X. (2019). Classiﬁed

prediction model of rockburst using rough sets-normal cloud. Neural

Computing and Applications, 31, 8185–8193.

Lu, A. H., Mao, X. B., & Liu, H. S. (2008). Physical simulation of rock

burst induced by stress waves. Journal of China University of Mining

and Technology, 18(3), 401–405.

Ma, T. H., Tang, C. A., Liu, F., Zhang, S. C., & Feng, Z. Q. (2021).

Microseismic monitoring, analysis and early warning of rockburst.

Geomatics, Natural Hazards and Risk, 12(1), 2956–2983.

Ma, T. H., Tang, C. A., Tang, L. X., Zhang, W. D., & Wang, L. (2015).

Rockburst characteristics and microseismic monitoring of deep-buried

tunnels for Jinping II Hydropower Station. Tunnelling and Under-

ground Space Technology, 49, 345–368.

Mark, C. (2016). Coal bursts in the deep longwall mines of the United

States. International Journal of Coal Science & Technology, 3(1), 1–9.

Merghadi, A., Yunus, A. P., Dou, J., Whiteley, J., ThaiPham, B., Bui, D.

T., ... Abderrahmane, B. (2020). Machine learning methods for

landslide susceptibility studies: A comparative overview of algorithm

performance. Earth-Science Reviews, 207, 103225.

96 J. Zhou et al. / Underground Space 14 (2024) 70–98

Meng, B., Jing, H. W., Chen, K. F., Su, H. J., Yang, S. Q., & Li, Y. H.

(2012). Shear and slip failure mechanism and control of tunnels with

weak surrounding rock. Chinese Journal of Geotechnical Engineering,

34(12), 2255–2262.

Meng, B., Jing, H., Chen, K., & Su, H. (2013). Failure mechanism and

stability control of a large section of very soft roadway surrounding

rock shear slip. International Journal of Mining Science and Technol-

ogy, 23(1), 127–134.

Meng, B., Jing, H., Yang, S., Wu, Y., Wang, Y., & Huang, Y. (2021).

Experimental investigation on shear behavior of intact sandstones

under constant normal stiﬀness conditions. International Journal of

Geomechanics, 21(2), 04020259.

Meng, Q. B., Kong, L. H., Wei, L. C., Shen, H. L., & Yang, Y. M. (2011).

Research Status and Prospect on The Support of Soft Rock Roadway

Engineering in Coal Mine. Coal, 01, 1–6 (in Chinese).

Min, Q., Lu, Y., Liu, Z., Su, C., & Wang, B. (2019). Machine learning

based digital twin framework for production optimization in petro-

chemical industry. International Journal of Information Management,

49, 502–519.

Mitri, H. S. (2007). Assessment of horizontal pillar burst in deep hard rock

mines. International Journal of Risk Assessment and Management, 7(5),

695–707.

¨ller, W. (1991). Numerical simulation of rock bursts. Mining Science

and Technology, 12(1), 27–42.

Oliver, L. C., Sampara, P., Pearson, D., Martell, J., & Zarnke, A. M.

(2022). Sarcoidosis in Northern Ontario hard-rock miners: A case

series. American Journal of Industrial Medicine, 65(4), 268–280.

Pan, J. F., Liu, S. H., Xia, Y. X., & Gao, J. M. (2021). Impact failure

characteristics and support method of roadway in broken coal seam

with large dip angle. Chinese Journal of Mining and Safety Engineering,

38(05), 946–953.

Pan, Y., & Zhang, L. (2021). Roles of artiﬁcial intelligence in construction

engineering and management: A critical review and future trends.

Automation in Construction, 122, 103517.

Pei, Q. T., Li, H. B., Liu, Y. Q., & Niu, J. T. (2013). Rockburst prediction

based on a modiﬁed grey evaluation model. Chinese Journal of Rock

Mechanics and Engineering, 32(10), 2088–2093.

Perol, T., Gharbi, M., & Denolle, M. (2018). Convolutional neural

network for earthquake detection and location. Science Advances, 4(2),

e1700578.

Pu, Y., Apel, D. B., Liu, V., & Mitri, H. (2019). Machine learning methods

for rockburst prediction-state-of-the-art review. International Journal

of Mining Science and Technology, 29(4), 565–570.

Qiao, C., Guo, Y. H., & Li, C. H. (2021). Study on rock burst prediction

of deep buried tunnel based on cusp catastrophe theory. Geotechnical

and Geological Engineering, 39(1), 1–15.

Qiu, S. L., Feng, X. T., Zhang, C. Q., & Wu, W. P. (2011). Development

and validation of rockburst vulnerability index (RVI) in deep hard

rock tunnels. Chinese Journal of Rock Mechanics and Engineering., 30

(6), 1126–1141 (in Chinese).

Qiu, Y. G., & Zhou, J. (2023). Short-term rockburst prediction in

underground project: Insights from an explainable and interpretable

ensemble learning model. Acta Geotechnica, 1–31.https://doi.org/

10.1007/s11440-023-01988-0.

Rehman, H., Naji, A. M., Ali, W., Junaid, M., Abdullah, R. A., & Yoo,

H. K. (2020). Numerical evaluation of new Austrian tunneling method

excavation sequences: A case study. International Journal of Mining

Science and Technology, 30(3), 381–386.

Rehman, H., Naji, A. M., Nam, K., Ahmad, S., Muhammad, K., & Yoo,

H. K. (2021). Impact of construction method and ground composition

on headrace tunnel stability in the Neelum-Jhelum Hydroelectric

Project: A case study review from Pakistan. Applied Sciences, 11(4),

1655.

Russenes, B. F. (1974). Analysis of rock spalling for tunnels in steep valley

sides [Master’s thesis, Norwegian Institute of Technology].

Shi, X. Z., Zhou, J., Dong, L., Hu, H. Y., Wang, H. Y., & Chen, S. R.

(2010). Application of unascertained measurement model to prediction

of classiﬁcation of rockburst intensity. Chinese Journal of Rock

Mechanics and Engineerin, 29(S1), 2720–2726 (in Chinese).

Shirani, F. R., & Taheri, A. (2019). Long-term prediction of rockburst

hazard in deep underground openings using three robust data mining

techniques. Engineering with Computers, 35(2), 659–675.

Simser, B. P. (2019). Rockburst management in Canadian hard rock

mines. Journal of Rock Mechanics and Geotechnical Engineering, 11(5),

1036–1043.

Singh, S. P. (1987). The inﬂuence of rock properties on the occurrence and

control of rockbursts. Mining Science and Technology, 5(1), 11–18.

Sun, Y., Li, G., Zhang, J., & Huang, J. (2021). Rockburst intensity

evaluation by a novel systematic and evolved approach: Machine

learning booster and application. Bulletin of engineering geology and

the environment, 80(11), 8385–8395.

Suorineni, F. T., Hebblewhite, B., & Saydam, S. (2014). Geomechanics

challenges of contemporary deep mining: A suggested model for

increasing future mining safety and productivity. Journal of the

Southern African Institute of Mining and Metallurgy, 114(12),

1023–1032.

Tajdus

´, A. (1997). Estimation of rockburst hazard basing on 3D stress

ﬁeld analysis. Ratio, 1(2), 3.

Tang, C. (1997). Numerical simulation of progressive rock failure and

associated seismicity. International Journal of Rock Mechanics and

Mining Sciences, 34(2), 249–261.

Tang, S., Shelden, D. R., Eastman, C. M., Pishdad-Bozorgi, P., & Gao, X.

(2019). A review of building information modeling (BIM) and the

internet of things (IoT) devices integration: Present status and future

trends. Automation in Construction, 101, 127–139.

Tan, Y. A., Guangzhong, S., & Zhi, G. U. O. (1991). A composite index

Krb criterion for the ejection characteristics of the burst rock. Chinese

Journal of Geology, 26(2), 193–200.

Tao, Z. Y. (1988). Support design of tunnels subjected to rockbursting pp.

ISRM-IS. ISRM International Symposium. ISRM.

Turchaninov, I. A., Markov, G. A., Gzovsky, M. V., Kazikayev, D. M.,

Frenze, U. K., Batugin, S. A., & Chabdarova, U. I. (1972). State of

stress in the upper part of the Earth’s crust based on direct

measurements in mines and on tectonophysical and seismological

studies. Physics of the Earth and Planetary Interiors, 6(4), 229–234.

Wang, B., Wang, W., Zhao, F., Fan & Tang (2014). Research on bolt

anchoring eﬀect based on self-supporting characteristics of roadway

surrounding rock. Geotechnical Mechanics, 35(7), 8.

Wang, H., Shi, R., Lu, C., Jiang, Y., Deng, D., & Zhang, D. (2019a).

Investigation of sudden faults instability induced by coal mining.

Safety Science, 115, 256–264.

Wang, L. S., Li, T. B., Xu, J., Xu, L. S., Jin, X. G., & Li, Y. L. (1999).

Rockburst and its intensity classiﬁcation of Erlangshan highway

tunnel. Highway, 28(2), 41–45.

Wang, M., Liu, Q., Wang, X., Shen, F., & Jin, J. (2020). Prediction of

rockburst based on multidimensional connection cloud model and set

pair analysis. International Journal of Geomechanics, 20(1), 04019147.

Wang, S. M., Zhou, J., Li, C. Q., Armaghani, D. J., Li, X. B., & Mitri, H.

S. (2021). Rockburst prediction in hard rock mines developing bagging

and boosting tree-based ensemble techniques. Journal of Central South

University, 28(2), 527–542.

Wang, S. Y., Lam, K. C., Au, S. K., Tang, C. A., Zhu, W. C., & Yang, T.

H. (2006). Analytical and numerical study on the pillar rockbursts

mechanism. Rock Mechanics and Rock Engineering, 39, 445–467.

Wang, X., Li, S., Xu, Z., Xue, Y., Hu, J., Li, Z., & Zhang, B. (2019b). An

interval fuzzy comprehensive assessment method for rock burst in

underground caverns and its engineering application. Bulletin of

Engineering Geology and the Environment, 78, 5161–5176.

Wu, M., Ye, Y., Wang, Q., & Hu, N. (2022). Development of rockburst

research: A comprehensive review. Applied Sciences, 12(3), 974.

Wu, Y. Z., Fu, Y. K., He, J., Chen, J. Y., Chu, X. W., & Meng, X. Z.

(2021). Principle and technology of ‘‘pressure relief-support-

protection”collaborative prevention and control in deep rock burst

roadway. Journal of China Mine Society, 46(1), 132–144.

Xi, Y. L., Xu, Y. Z., & Wang, L. C. (2017). Study and countermeasures on

key technologies for long deeply buried tunnel of Qireha Starr

Hydropower Station. Water Resources and Hydropower Engineering,

48(10), 26–30.

Xu, Y. N., Xu, W. S., Wang, Y. H., Tham, L. G., & Lee, P. K. K. (2002).

Simulation testing and mechanism studies on rockburst. Chinese

Journal of Rock Mechanics and Engineering, 21, 1462–1466.

Yang, Z. Y., Liu, C., Zhu, H. Z., Xie, F. X., Dou, L. M., & Chen, J. H.

(2019). Mechanism of rock burst caused by fracture of key strata

during irregular working face mining and its prevention methods.

Journal of Mining Science and Technology: English Edition, 29(6), 9.

Yang, S. Q. (1993). An experimental study on rockburst mechanism

around tunnels by physical simulation. Journal of Wuhan University of

Hydraulic and Electric Engineering, 26(2), 160–166.

Yin, X., Liu, Q., Pan, Y., Huang, X., Wu, J., & Wang, X. (2021). Strength

of stacking technique of ensemble learning in rockburst prediction with

J. Zhou et al. / Underground Space 14 (2024) 70–98 97

imbalanced data: Comparison of eight single and ensemble models.

Natural Resources Research, 30, 1795–1815.

Yu, S., & Ma, J. (2021). Deep learning for geophysics: Current and future

trends. Reviews of Geophysics, 59(3), e2021RG000742.

Zhang, C. Q., Lu, J. J., Chen, J., Zhou, H., & Yang (2017). Discussion on

rock burst proneness indexes and their relation. Rock and Soil

Mechanics, 38(5), 1397–1404.

Zhang, C. Q., Zhou, H., & Feng, X. T. (2011). An index for estimating the

stability of brittle surrounding rock mass: FAI and its engineering

application. Rock Mechanics and Rock Engineering, 44(4), 401–414.

Zhang, G., Pan, Y., Zhang, L., & Tiong, R. L. K. (2020a). Cross-scale

generative adversarial network for crowd density estimation from

images. Engineering Applications of Artiﬁcial Intelligence, 94, 103777.

Zhang, J., Wang, Y., Sun, Y., & Li, G. (2020b). Strength of ensemble

learning in multiclass classiﬁcation of rockburst intensity. International

Journal for Numerical and Analytical Methods in Geomechanics, 44(13),

1833–1853.

Zhang, L., Zhang, X., Wu, J., Zhao, D., & Fu, H. (2020c). Rockburst

prediction model based on comprehensive weight and extension

methods and its engineering application. Bulletin of Engineering

Geology and the Environment, 79, 4891–4903.

Zhao, H., Chen, B., & Zhu, C. (2021). Decision tree model for rockburst

prediction based on microseismic monitoring. Advances in Civil

Engineering, 2021, 1–14.

Zhou, J. (2015a). Supervised learning methods for strain-based rockburst

prediction and crater depth estimation. Ph.D thesis, Changsha, Central

South University.

Zhou, J., Chen, C., Wang, M., & Khandelwal, M. (2021a). Proposing a

novel comprehensive evaluation model for the coal burst liability in

underground coal mines considering uncertainty factors. International

Journal of Mining Science and Technology, 31(5), 99–812.

Zhou, J., Chen, C., Wei, C., & Du, K. (2022a). An Improved Connection

Cloud Model of an Updated Database: A Multicriteria Uncertainty

Model for Coal Burst Liability Evaluation. Natural Resources

Research, 31(3), 1687–1704.

Zhou, J., Koopialipoor, M., Li, E., & Armaghani, D. J. (2020). Prediction

of rockburst risk in underground projects developing a neuro-bee

intelligent system. Bulletin of Engineering Geology and the Environment,

79(8), 4265–4279.

Zhou, J., Li, X. B., & Mitri, H. S. (2015b). Comparative performance of

six supervised learning methods for the development of models of hard

rock pillar stability prediction. Natural Hazards, 79(1), 291–316.

Zhou, J., Li, X., & Mitri, H. S. (2016a). Classiﬁcation of rockburst in

underground projects: Comparison of ten supervised learning meth-

ods. Journal of Computing in Civil Engineering, 30(5), 04016003.

Zhou, J., Li, X. B., & Mitri, H. S. (2018). Evaluation method of rockburst:

State-of-the-art literature review. Tunnelling and Underground Space

Technology, 81, 632–659.

Zhou, J., Li, X. B., & Shi, X. Z. (2012). Long-term prediction model of

rockburst in underground openings using heuristic algorithms and

support vector machines. Safety Science, 50(4), 629–644.

Zhou, J., Guo, H., Koopialipoor, M., Jahed Armaghani, D., & Tahir, M.

M. (2021b). Investigating the eﬀective parameters on the risk levels of

rockburst phenomena by developing a hybrid heuristic algorithm.

Engineering with Computers, 37(3), 1679–1694.

Zhou, J., Shen, X., Qiu, Y., Li, E., Rao, D., & Shi, X. (2021c). Improving

the eﬃciency of microseismic source locating using a heuristic

algorithm-based virtual ﬁeld optimization method. Geomechanics and

Geophysics for Geo-Energy and Geo-Resources, 7(3), 1–18.

Zhou, J., Shen, X., Qiu, Y., Shi, X., & Khandelwal, M. (2022b). Cross-

correlation stacking-based microseismic source location using three

metaheuristic optimization algorithms. Tunnelling and Underground

Space Technology, 126, 104570.

Zhou, J., Shi, X. Z., Dong, L., Hu, H. Y., & Wang, H. Y. (2010). Fisher

discriminant analysis model and its application for prediction of

classiﬁcation of rockburst in deep-buried long tunnel. Journal of Coal

Science and Engineering (China), 16(2), 144–149.

Zhou, J., Shi, X. Z., Huang, R. D., Qiu, X. Y., & Chen, C. (2016b).

Feasibility of stochastic gradient boosting approach for predicting

rockburst damage in burst-prone mines. Transactions of Nonferrous

Metals Society of China, 26(7), 1938–1945.

Zhu, Q., Zhao, X., & Westman, E. (2021). Review of the evolution of

mining-induced stress and the failure characteristics of surrounding

rock based on microseismic tomography. Shock and Vibration, 2021,

1–19.

Zhu, S., Feng, Y., Jiang, F., & Liu, J. (2018). Mechanism and risk

assessment of overall-instability-induced rockbursts in deep island

longwall panels. International Journal of Rock Mechanics and Mining

Sciences, 106, 342–349.

Zhu, W. C., Li, Z. H., Zhu, L., & Tang, C. A. (2010). Numerical

simulation on rockburst of underground opening triggered by dynamic

disturbance. Tunnelling and Underground Space Technology, 25(5),

587–599.

Zubelewicz, A., & Mroz, Z. (1983). Numerical simulation of rock burst

processes treated as problems of dynamic instability. Rock Mechanics

and Rock Engineering, 16(4), 253–274.

Wen, C.P., 2008. Application of attribute synthetic evaluation system in

prediction of possibility and classiﬁcation of rockburst. Engineering

Mechanics. 25 (6), 153–158. (in Chinese)

98 J. Zhou et al. / Underground Space 14 (2024) 70–98

Predicting the magnitude of technogenic earthquakes during underground mining of the Zhezkazgan ore field

Article

Mar 2024

Purpose. Determining a relationship between the shear/failure of the rock mass and the technogenic earthquakes caused by them during underground mining of ore deposits with the derivation of an equation for the dependence of the technogenic earthquake indicators for the conditions of the Zhezkazgan copper-ore field. Methods. The research methodology consists of studying and analyzing early research on technogenic earthquakes by the method of statistical data processing. Calculation of rock mass deformation distribution in the study area is based on numerical modeling in the Comsol Multiphysics 5.6 and MATLAB 2020 software package environment. Calculations are performed based on solving a plane problem using the Finite Element Method (FEM). Findings. Based on the transition from the failure area to subsidence and shear values of the overlying rock mass stratum, an equation for the dependence of the earthquake magnitude on the numerical values of the mass subsidence or shear has been obtained. Originality. For the first time, based on the physics and geomechanics of rock mass shear processes, empirical-analytical formulas have been obtained that make it possible to predict the technogenic earthquake magnitude during underground mining of ore deposits in the conditions of the Zhezkazgan copper-ore field. Practical implications. Preliminary predictive calculations made by the obtained formulas for the conditions of the active mine No. 31 of the East-Zhezkazgan mine, ТОО Kazakhmys Smelting, show acceptable results of magnitude value, comparable to in-situ measurements during field mining. This prediction makes it possible to pre-calculate the technogenic earthquake magnitude at the stage of designing mining operations and make appropriate scientifically sound decisions during further mining of the field.

Classifying rockburst with confidence: A novel conformal prediction approach

Article

Full-text available

Jan 2024

The scientific community recognizes the seriousness of rockbursts and the need for effective mitigation measures. The literature reports various successful applications of machine learning (ML) models for rockburst assessment; however, a significant question remains unanswered: How reliable are these models , and at what confidence level are classifications made? Typically, ML models output single rockburst grade even in the face of intricate and out-of-distribution samples, without any associated confidence value. Given the susceptibility of ML models to errors, it becomes imperative to quantify their uncertainty to prevent consequential failures. To address this issue, we propose a conformal prediction (CP) framework built on traditional ML models (extreme gradient boosting and random forest) to generate valid classifications of rockburst while producing a measure of confidence for its output. The proposed framework guarantees marginal coverage and, in most cases, conditional coverage on the test dataset. The CP was evaluated on a rockburst case in the Sanshandao Gold Mine in China, where it achieved high coverage and efficiency at applicable confidence levels. Significantly, the CP identified several ''confident" classifications from the traditional ML model as unreliable, necessitating expert verification for informed decision-making. The proposed framework improves the reliability and accuracy of rockburst assessments, with the potential to bolster user confidence.

Occurrence mechanism and prevention technology of rockburst, coal bump and mine earthquake in deep mining

Article

Full-text available

May 2024

Rockburst, coal bump, and mine earthquake are the most important dynamic disaster phenomena in deep mining. This paper summarizes the differences and connections between rockburst, coal bumps and mine earthquakes in terms of definition, mechanism, phenomenon, evaluation index, etc. The definition and evolution progress of the three disaster categories are summarized, as well as the monitoring, early warning, and prevention measures are also presented. Firstly, by combining theoretical research with specific technologies and engineering field cases, the main categories and failure mechanisms of rockburst, coal bumps and mine earthquakes are introduced. Then, the evaluation indexes of coal bump and mine earthquake are summarized, and a new evaluation index of rockburst is given. Finally, the characteristics of monitoring, early warning technologies and prevention methods of rockburst, coal bumps, and mine earthquakes are discussed in technology and application. At last, the future directions of rockburst, coal bump and mine earthquake are put forward.

A Visual Survey of Tunnel Boring Machine (TBM) Performance in Tunneling Excavation: Mainstream Direction, Brief Review and Future Prospects

Article

Full-text available

May 2024

This study employs scientometric analysis to investigate the current trajectory of research on tunnel boring machine (TBM) performance and collaborative efforts. Utilizing software tools like Pajek 5.16 and VOSviewer 1.6.18, it scrutinizes literature from 2000 to 2021 sourced from the Web of Science (WOS). The findings illuminate TBM research as an interdisciplinary and intersectoral field attracting increasing national and institutional attention. Notable contributions from China, Iran, the United States, Turkey, and Australia underscore the global significance of TBM research. The recent upsurge in annual publications, primarily driven by Chinese research initiatives, reflects a renewed vigor in TBM exploration. Additionally, the paper presents a succinct evaluation of TBM advantages and drawbacks compared to conventional drill and blast methods, discussing key considerations in excavation methodology selection. Moreover, the study comprehensively reviews TBM performance prediction models, categorizing them into theoretical, empirical, and artificial intelligence-driven approaches. Finally, rooted in metaverse theory, the discourse delves into the immersive learning model and the architecture of a TBM metaverse. In the future, the immersive training and learning model diagram can be employed in scenarios such as employee training and the promotion of safety knowledge. Additionally, the TBM metaverse architecture can simulate, monitor, diagnose, predict, and control the organization, management, and service processes and behaviors of TBMs. This will enhance efficient collaboration across various aspects of the project production cycle. This forward-looking perspective anticipates future trends in TBM technology, emphasizing societal impact and enhancement of economic benefits.

Predicting Rock Burst in Underground Engineering Leveraging a Novel Metaheuristic-Based LightGBM Model

Article

Full-text available

Apr 2024
Comput Model Eng Sci

Rock bursts represent a formidable challenge in underground engineering, posing substantial risks to both infrastructure and human safety. These sudden and violent failures of rock masses are characterized by the rapid release of accumulated stress within the rock, leading to severe seismic events and structural damage. Therefore, the development of reliable prediction models for rock bursts is paramount to mitigating these hazards. This study aims to propose a tree-based model—a Light Gradient Boosting Machine (LightGBM)—to predict the intensity of rock bursts in underground engineering. 322 actual rock burst cases are collected to constitute an exhaustive rock burst dataset, which serves to train the LightGBM model. Two population-based metaheuristic algorithms are used to optimize the hyperparameters of the LightGBM model. Finally, the sensitivity analysis is used to identify the predominant factors that may incur the occurrence of rock bursts. The results show that the population-based metaheuristic algorithms have a good ability to search out the optimal hyperparameters of the LightGBM model. The developed LightGBM model yields promising performance in predicting the intensity of rock bursts, with which accuracy on training and testing sets are 0.972 and 0.944, respectively. The sensitivity analysis discloses that the risk of occurring rock burst is significantly sensitive to three factors: uniaxial compressive strength (σc), stress concentration factor (SCF), and elastic strain energy index (Wet). Moreover, this study clarifies the particular impact of these three factors on the intensity of rock bursts through the partial dependence plot.

Risk Assessment and Analysis of Rock Burst under High-Temperature Liquid Nitrogen Cooling

Article

Full-text available

Feb 2024

Rock burst, an important kind of geological disaster, often occurs in underground construction. Rock burst risk assessment, as an important part of engineering risk assessment, cannot be ignored. Liquid nitrogen fracturing is a new technology used in the geological, oil, and gas industries to enhance productivity. It involves injecting liquid nitrogen into reservoir rocks to induce fractures and increase permeability, effectively reducing rock burst occurrences and facilitating the flow of oil or gas toward the wellbore. The research on rock burst risk assessment technology is the basis of reducing rock burst geological disasters, which has important theoretical and practical significance. This article examines the temperature treatment of two types of rocks at 25 °C, 100 °C, 200 °C, 300 °C, and 400 °C, followed by immersion in a liquid nitrogen tank. The temperature difference between the liquid nitrogen and the rocks may trigger rock bursting. The research focused on analyzing various characteristics of rock samples when exposed to liquid nitrogen. This included studying the stress–strain curve, elastic modulus, strength, cross-section analysis, wave velocity, and other relevant aspects. Under the influence of high temperature and a liquid nitrogen jet, the wave velocity of rocks often changes. The structural characteristics and possible hidden dangers of rocks can be understood more comprehensively through section scanning analysis. The stress–strain curve describes the deformation and failure behavior of rocks under different stress levels, which can help to evaluate their stability and structural performance. The investigation specifically focused on the behavior of rocks subjected to high temperatures and liquid nitrogen. By analyzing the stress–strain curves, researchers were able to identify the precursors and deformation processes that occur before significant deformation or failure. These findings have implications for the mechanical properties and stability of the rocks.

Study on the effect of multiple optimization algorithms on rockburst prediction models

Preprint

Full-text available

Feb 2024

The rapidly expanding area of rockburst prediction has drawn a lot of interest because of its enormous potential to lower the risk of engineering disasters, enhance mine production safety, and protect employee lives. Consequently, the goal of this research is to forecast the rockburst intensity class for the prediction objective by optimizing four single machine learning models (SVM, DT, CNN, and RF) utilizing fifteen optimization algorithms (Bayes, SSA, DBO, SCA, SA, PSO, SO, POA, GWO, IGWO, AVOA, CSA, GTO, NGO, and WSO). The hybrid models were trained using a ten-fold cross-validation, and each hybrid model's performance was examined statistically. The SMOTE method then oversampled the original dataset in order to examine how the data equalization issue affected the hybrid models. The findings demonstrate that, in the original dataset, all optimization strategies increase the accuracy of the DT, CNN, and RF models; however, the balanced original dataset has a greater impact on the SVM models. And once the dataset is balanced, every optimization algorithm improves the accuracy of the SVM model and decreases the accuracy of the DT model; however, for the CNN and RF models, the majority of optimization algorithms improve the accuracy while only a small percentage of them do the opposite. An essential reference value for the development of later rock burst prediction models is provided by this study.

Novel rockburst prediction criterion with enhanced explainability employing CatBoost and nature-inspired metaheuristic technique

Article

Jun 2024

Toward Precise Long-Term Rockburst Forecasting: A Fusion of SVM and Cutting-Edge Meta-heuristic Algorithms

Article

Jun 2024

Rockburst is one of the most hazardous geological disasters in underground engineering due to its complex causes and destructive nature. To address this, there is an imperative for methodologies that can predict rockbursts quickly and effectively to mitigate preemptively the risks and damages. In this study, 259 rockburst instances were analyzed, employing six rockburst feature parameters: maximum tangential stress (σθ), uniaxial compressive strength of rock (σc), uniaxial tensile strength of rock (σt), stress coefficient (σθ/σt), rock brittleness coefficient (σc/σt), and elastic energy index (Wet) as inputs. By integrating three novel meta-heuristic algorithms—dingo optimization algorithm (DOA), osprey optimization algorithm (OOA), and rime-ice optimization algorithm (RIME)—with support vector machine (SVM), hybrid models for long-term rockburst trend prediction were constructed. Performance evaluations through fivefold cross-validation revealed that for the no rockbursts, DOA–SVM (Pop = 200) demonstrated superior predictive performance, achieving an accuracy of 0.9808, precision of 0.9231, recall of 1, and an F1-score of 0.96. For moderate rockbursts, OOA–SVM (Pop = 100) emerged as the most effective, registering an accuracy of 0.9808, precision of 0.9545, recall of 1, and an F1-score of 0.9767. For light and severe rockbursts, DOA–SVM, OOA–SVM, and RIME–SVM showcased comparable predictive outcomes. However, these hybrid models outperformed traditional SVM models optimized with conventional algorithms in terms of accuracy across all rockburst hazard levels. Moreover, the hybrid models underwent additional validation with a new dataset of 20 rockburst instances collected globally, confirming their robust efficacy and exceptional generalization capabilities. An ensuing analysis using local interpretable model-agnostic explanations (LIME) on the six key feature parameters revealed a significant positive correlation between σθ and Wet with the severity of rockbursts. These results not only affirm the superior optimization performance of the DOA, OOA, and RIME algorithms but also their substantial potential to enhance the predictive accuracy of machine learning models in forecasting long-term rockbursts.

State-of-the-art review of machine learning and optimization algorithms applications in environmental effects of blasting

Article

Full-text available

Jan 2024
ARTIF INTELL REV

The technological difficulties related with blasting operations have become increasingly significant. It is crucial to give due consideration to the evaluation of rock fragmentation and the threats posed by environmental effect of blasting (EEB). To address these challenges, numerous scholars have conducted extensive research employing various assessment techniques with the aim of mitigating risks and preventing the emergence of unfavorable EEB. The occurrence of EEB is prevalent during the excavation of hard rock, and it presents significant hazards to personnel safety, equipment integrity, and operational continuity. Therefore, conducting a systematic review of EEB is of utmost importance as it enables a comprehensive understanding of the contributing factors. Such an understanding plays a vital role in advancing EEB prediction and prevention methods. The careful selection of an appropriate EEB assessment method is a crucial aspect of blasting operations. However, there is a lack of comprehensive discussions on the applications of machine learning (ML) and optimization algorithms (OA) in addressing various EEB. Only a limited number of papers have briefly touched upon this topic. Therefore, the primary objective of this paper is to bridge this gap by conducting an analysis of global trends using CiteSpace and VOSviewer software from the year 2000 onwards. It comprehensively explores EEB classification and definition, encompassing air overpressure (AOp), ground vibration, dust, backbreak, flyrock, and rock fragmentation. Furthermore, the paper provides a compendium of the most recent ML and OA prediction techniques used to addresses EEB. Finally, the paper concludes by proposing future directions for exploring innovative approaches that combine data-driven ML techniques with knowledge-based or physics-based methods. Such integration has the potential to mitigate hazards during blasting operations and reduce the likelihood of unfavorable EEB occurrences.

Short-term rockburst prediction in underground project: insights from an explainable and interpretable ensemble learning model

Article

Full-text available

Jul 2023

Rockburst is a frequent challenge during tunnel and other underground construction and is an extreme rock damage phenomenon. Therefore, it is very crucial to accurately estimate the damage potential of rockburst events. Microseismic (MS) monitoring can be used to obtain the relevant MS parameters for short-term rockburst prediction in real time that reflect the evolution of short-term rockburst. In this study, short-term rockburst potential data containing 7 MS parameters (cumulative number of events, cumulative released energy, cumulative apparent volume, event rate, energy rate, apparent volume rate, and incubation time) and 91 rockburst events (none rockburst, low rockburst, moderate rockburst, and high rockburst) were collected from the Jinping Hydropower Station diversion tunnel project in China. The objective of this paper is to propose an ensemble learning (EML) model based on the LévyFlight-Jaya optimization (LFJaya) and fivefold cross-validation (CV) method to achieve an accurate prediction of short-term rockburst damage potential using MS information. The EML consists of light gradient boosting machine (LGBM), extreme gradient boosting (XGBoost), random forest (RF), support vector machine (SVM), and logistic regression (LR), with seven MS parameters as the EML inputs and four rockburst levels as target variables. 70% and 30% of the cases were randomly selected for training and testing, respectively. Five metrics (accuracy, kappa, precision, recall, and F1-score) and nonparametric statistical tests were used to evaluate the performance of the model. It can be observed from the results of this study that the proposed EML has a higher test accuracy (89.29%) than the multiple base classifiers used in the study. With the use of the ensemble model, the decision boundary becomes more precise and overfitting is significantly improved. Additionally, the internal decision-making process of EML was elucidated through an analysis of the model parameters using SHapley Additive exPlanations (SHAP) and Local Interpretable Model-agnostic Explanations (LIME). It was discovered that the cumulative released energy, the number of MS events, and the cumulative apparent volume (which reflects the number and strength of rock fractures) exert a significant influence on the prediction of short-term rockburst potential. Finally, developed graphical user interface (GUI) accurately predicted six instances of rockburst in the deeply buried tunnel of Jinping. Verification results indicated that the proposed EML exhibits strong generalization and can effectively utilize MS information to achieve precise short-term rockburst potential predictions.

Development of Rockburst Research: A Comprehensive Review

Article

Full-text available

Jan 2022

Rockburst is a difficult problem in underground resource development and infrastructure construction. With scientific and technological progress, more research technologies and methods to prevent rockburst have been proposed and applied. Research content has become more comprehensive, and positive progress and achievements have been made. However, due to the complexity of rockburst control factors and the sudden occurrence of rockburst, the occurrence mechanism and accurate prediction of occurrence intensity and time are still difficult worldwide problems. In this paper, the research development and achievements of rockburst are reviewed. Firstly, various definitions and types of rockburst are briefly summarized. Then, research progress and achievements in four aspects are comprehensively reviewed, including methods and techniques of rockburst research (theoretical research, numerical simulation, physical model tests and in situ monitoring), mechanism of rockburst, classification and prediction of rockburst (empirical criteria, in situ monitoring, mathematical model approaches and rockburst chart) and prevention and control of rockburst. Of particular significance is that the classification and prediction with prevention and control are summarized in detail. Finally, limitations, deficiencies and some promising directions for future research are listed.

An Improved Connection Cloud Model of an Updated Database: A Multicriteria Uncertainty Model for Coal Burst Liability Evaluation

Article

Full-text available

Mar 2022

In the context of complex geological environments, coal burst liability (CBL) evaluation struggles with uncertainty and fuzziness of indices and with obtaining satisfactory accuracy. Therefore, an improved connection cloud model (ICCM) is proposed for CBL evaluation based on set pair analysis referring to four indices. The numerical characteristics of the ICCM, namely expectation (Ex), entropy (En), hyper-entropy (He) and parameter (f), were determined by set pair analysis and classification standards. Then, the index measurement-based entropy weight method and the analysis hierarchy process (AHP) were used to measure clearly the importance of each index. Finally, Euclidean distance was employed to classify the CBL objectively, making full use of the valuable information mined from the data. An updated CBL database (of 147 data records) was constructed to validate the potential of the proposed algorithm for CBL classification. Additionally, a graphical user interface was developed to make the CBL evaluation visible. The classification metrics of the ICCM, i.e., accuracy and Kappa coefficient, were 0.88 and 0.772, respectively, which were better than those of the other conventional uncertainty methods. The results reveal that the proposed ICCM is a prospective tool for uncertainty information treatment and subjectivity removal in the field of coal burst control.

Novel Ensemble Tree Solution for Rockburst Prediction Using Deep Forest

Article

Full-text available

Mar 2022

The occurrence of rockburst can cause significant disasters in underground rock engineering. It is crucial to predict and prevent rockburst in deep tunnels and mines. In this paper, the deficiencies of ensemble learning algorithms in rockburst prediction were investigated. Aiming at these shortages, a novel machine learning model, deep forest, was proposed to predict rockburst risk. The deep forest combines the characteristics of deep learning and ensemble models, which can solve complex problems. To develop the deep forest model for rockburst prediction, 329 real rockburst cases were collected to build a comprehensive database for intelligent analysis. Bayesian optimization was proposed to tune the hyperparameters of the deep forest. As a result, the deep forest model achieved 100% training accuracy and 92.4% testing accuracy, and it has more outstanding capability to forecast rockburst disasters compared to other widely used models (i.e., random forest, boosting tree models, neural network, support vector machine, etc.). The results of sensitivity analysis revealed the impact of variables on rockburst levels and the applicability of deep forest with a few input parameters. Eventually, real cases of rockburst in two gold mines, China, were used for validation purposes while the needed data sets were prepared by field observations and laboratory tests. The promoting results of the developed model during the validation phase confirm that it can be used with a high level of accuracy by practicing engineers for predicting rockburst occurrences.

Sarcoidosis in Northern Ontario hard‐rock miners: A case series

Article

Full-text available

Feb 2022

Sarcoidosis is a rare multisystem granulomatous disease traditionally considered to be of unknown etiology. The notion that sarcoidosis has no known cause is called into question with the increasing number of case reports and epidemiologic studies showing associations between occupational exposures and disease published in the past 10–20 years. Occupational exposures for which associations are strongest and most consistent are silica and other inorganic dusts, World Trade Center (WTC) dust, and metals. Occupations identified as at‐risk for sarcoidosis include construction workers; iron‐foundry and diatomaceous earth workers; WTC emergency responders; and metal workers. We report here 12 cases of sarcoidosis in a cohort of hard‐rock miners in Northern Ontario, Canada. To our knowledge sarcoidosis has not been reported previously in hard‐rock miners. The cases are all male and Caucasian, with average age 74 years. At the time of diagnosis, two were never smokers; six, former smokers; and four, current smokers. Five have extrapulmonary sarcoidosis: two cardiac and three endocrine (hypercalciuria). Using occupational histories and air sampling data from the gold, uranium, and base‐metal mines in which they worked, we examined exposure of each case to respirable crystalline silica (RCS). The annual mean RCS exposure for the 12 cases was 0.14 mg/m3 (range: 0.06–1.3 mg/m3); and the mean cumulative RCS exposure was 1.93 mg/m3 years (range: 0.64–4.03 mg/m3 years). We also considered their exposure to McIntyre Powder, an aluminum powder used for silicosis prophylaxis.

Novel ensemble intelligence methodologies for rockburst assessment in complex and variable environments

Article

Full-text available

Feb 2022

Rockburst is a severe geological hazard that restricts deep mine operations and tunnel constructions. To overcome the shortcomings of widely used algorithms in rockburst prediction, this study investigates the ensemble trees, i.e., random forest (RF), extremely randomized tree (ET), adaptive boosting machine (AdaBoost), gradient boosting machine, extreme gradient boosting machine (XGBoost), light gradient boosting machine, and category gradient boosting machine, for rockburst estimation based on 314 real rockburst cases. Additionally, Bayesian optimization is utilized to optimize these ensemble trees. To improve performance, three combination strategies, voting, bagging, and stacking, are adopted to combine multiple models according to training accuracy. ET and XGBoost receive the best capabilities (85.71% testing accuracy) in single models, and except for AdaBoost, six ensemble trees have high accuracy and can effectively foretell strong rockburst to prevent large-scale underground disasters. The combination models generated by voting, bagging, and stacking perform better than single models, and the voting 2 model that combines XGBoost, ET, and RF with simple soft voting, is the most outstanding (88.89% testing accuracy). The performed sensitivity analysis confirms that the voting 2 model has better robustness than single models and has remarkable adaptation and superiority when input parameters vary or miss, and it has more power to deal with complex and variable engineering environments. Eventually, the rockburst cases in Sanshandao Gold Mine, China, were investigated, and these data verify the practicability of voting 2 in field rockburst prediction.

A new empirical chart for coal burst liability classification using Kriging method

Article

May 2023

Coal burst is a catastrophic event that induced by a large variety of certainty and uncertainty factors, and many methods have been proposed to evaluate the risk of this hazard. Conventional evaluation models or empirical criteria are influenced by the complex modelling process or undesirable accuracy. In this study, a total of 147 groups of coal burst records were used to establish the empirical classification model based on elastic energy index (Wet) and impact energy index (Kc). The classification boundaries of coal burst liabilities (CBLs), which was fitted to quantitatively analyze the risk level, and its distribution characteristics are displayed on 2D chart using Kriging method. Additionally, 43 groups of test samples were collected to further validate the reliability of the constructed spatial interpolation model. The results revealed that the classification performance of Kriging model outperforms other uncertainty-based method with accuracy 91%. It can be a valuable and helpful tool for designers to conduct the geological hazard prevention and initial design.

Rockburst prediction in hard rock mines developing bagging and boosting tree-based ensemble techniques

Article

Sep 2020

Rockburst prediction is of vital significance to the design and construction of underground hard rock mines. A rockburst database consisting of 102 case histories, i.e., 1998–2011 period data from 14 hard rock mines was examined for rockburst prediction in burst-prone mines by three tree-based ensemble methods. The dataset was examined with six widely accepted indices which are: the maximum tangential stress around the excavation boundary (MTS), uniaxial compressive strength (UCS) and uniaxial tensile strength (UTS) of the intact rock, stress concentration factor (SCF), rock brittleness index (BI), and strain energy storage index (EEI). Two boosting (AdaBoost.M1, SAMME) and bagging algorithms with classification trees as baseline classifier on ability to learn rockburst were evaluated. The available dataset was randomly divided into training set (2/3 of whole datasets) and testing set (the remaining datasets). Repeated 10-fold cross validation (CV) was applied as the validation method for tuning the hyper-parameters. The margin analysis and the variable relative importance were employed to analyze some characteristics of the ensembles. According to 10-fold CV, the accuracy analysis of rockburst dataset demonstrated that the best prediction method for the potential of rockburst is bagging when compared to AdaBoost.M1, SAMME algorithms and empirical criteria methods.

Cross-correlation stacking-based microseismic source location using three metaheuristic optimization algorithms

Article

Aug 2022
TUNN UNDERGR SP TECH

Microseismic location systems tend to be high-speed and precise. However, the requirement of high precision tends to slow down the calculation speed. Fortunately, metaheuristics are able to alleviate this problem. In this research, metaheuristic algorithms are used to improve the performance of cross-correlation stacking (CCS). CCS has able to provide excellent location accuracy as it uses more information in the entire waveform for location. However, this method often requires more calculation time due to its complex mathematical modeling. To overcome this problem, various metaheuristic algorithms (i.e. moth flame optimization (MFO), ant lion optimization (ALO) and grey wolf optimization (GWO)) have been used to improve CCS. It has been found that appropriate control parameters can improve the metaheuristic algorithm performance manyfold. So, these control parameters have been adjusted based on three different perspectives, i.e. success rate (SR), computational efficiency and convergence performance. The results show that these models are able to provide better location efficiency compared to the full grid search (FGS) and particle swarm optimization (PSO) based on ensuring good location accuracy. It is also found that MFO is significantly better than the other metaheuristic algorithms. In addition, the superiority of CCS over traditional location methods is verified through comprehensive tests, and the influence of the speed model and the number of sensors on the location performance of CCS was tested.

Intelligent rockburst prediction model with sample category balance using feedforward neural network and Bayesian optimization

Article

Feb 2022

The rockburst prediction becomes more and more challenging due to the development of deep underground projects and constructions. Increasing numbers of intelligent algorithms are used to predict and prevent rockburst. This paper investigated the drawbacks of neural networks in rockburst prediction, and aimed at these shortcomings, Bayesian optimization and the synthetic minority oversampling technique+Tomek Link (SMOTETomek) were applied to efficiently develop the feedforward neural network (FNN) model for rockburst prediction. In this regard, 314 real rockburst cases were collected to establish a database for modeling. The database was divided into a training set (80%) and a test set (20%). The maximum tangential stress, uniaxial compressive strength, tensile strength, stress ratio, brittleness ratio, and elastic strain energy were selected as input parameters. Bayesian optimization was implemented to find the optimal hyperparameters in FNN. To eliminate the effects of imbalanced category, SMOTETomek was adopted to process the training set to obtain a balanced training set. The FNN developed by the balanced training set received 90.48% accuracy in the test set, and the accuracy improved 12.7% compared to the imbalanced training set. For interpreting the FNN model, the permutation importance algorithm was introduced to analyze the relative importance of input variables. The elastic strain energy was the most essential variable, and some measures were proposed to prevent rockburst. To validate the practicability, the FNN developed by the balanced training set was utilized to predict rockburst in Sanshandao Gold Mine, China, and it had outstanding performance (accuracy 100%).

Rockburst prediction and prevention in underground space excavation

Abstract

Recommended publications

A rockburst prediction model based on extreme learning machine with improved Harris Hawks optimizati...

Ensemble Tree Model for Long-Term Rockburst Prediction in Incomplete Datasets

Development of Rockburst Research: A Comprehensive Review

Measurement and Classification Criteria of Strength Decrease Rate and Brittleness Indicator Index fo...