Recent Advances and Trends of Cyber-Physical Systems and Big Data Analytics in Industrial Informatics

Abstract

In today's competitive business environment, companies are facing challenges in dealing with big data issues for rapid decision making for improved productivity. Many manufacturing systems are not ready to manage big data due to the lack of smart analytics tools. Germany is leading a transformation toward 4th Generation Industrial Revolution (Industry 4.0) based on Cyber-Physical System based manufacturing and service innovation. As more software and embedded intelligence are integrated in industrial products and systems, predictive technologies can further intertwine intelligent algorithms with electronics and tether-free intelligence to predict product performance degradation and autonomously manage and optimize product service needs. This article addresses the trends of industrial transformation in big data environment as well as the readiness of smart predictive informatics tools to manage big data to achieve transparency and productivity.

Full text

Proceeding of Int. Conference on Industrial Informatics (INDIN) 2014
Recent Advances and Trends of Cyber-Physical
Systems and Big Data Analytics in Industrial
Informatics
Jay Lee, Behrad Bagheri, Hung-An Kao
Center for Intelligent Maintenance Systems
University of Cincinnati
Cincinnati, United States
Abstract— In today’s competitive business environment,
companies are facing challenges in dealing with big data issues
for rapid decision making for improved productivity. Many
manufacturing systems are not ready to manage big data due to
the lack of smart analytics tools. Germany is leading a
transformation toward 4th Generation Industrial Revolution
(Industry 4.0) based on Cyber-Physical System based
manufacturing and service innovation. As more software and
embedded intelligence are integrated in industrial products and
systems, predictive technologies can further intertwine intelligent
algorithms with electronics and tether-free intelligence to predict
product performance degradation and autonomously manage
and optimize product service needs. This article addresses the
trends of industrial transformation in big data environment as
well as the readiness of smart predictive informatics tools to
manage big data to achieve transparency and productivity.
Keywords—Industry 4.0; Cyber Physical Systems; Prognostics
and Health Management; Big Data;
I. INTRODUCTION
In the past decade, emergence of promising tools such as
Enterprise Systems provided companies with solutions to
improve their productivity and service quality. But today's
competitive nature of world industry enforces companies to
implement more recent technologies to secure their position
among competitors [1]. Recent technological advances in the
field of communication and computer science have provided
cost-effective solutions for companies to acquire and transfer
gigantic amount of data from their fleet of assets. Consequently
Handling these huge sets of data is not easily achievable,
therefore supporting "Big Data" is one of the most recent topics
in the world industry. Introduction of methods and terms such
as internet of things and interconnected systems are among the
efforts of researchers and industrial companies to address
applicable solutions in the "Big Data" environment. Therefore,
the requirement of systematic approaches to handle and
analyze enterprise data in the Big Data environment is the
purpose of several research studies including current paper. [2].
II. RECENT ADVANCES AND TERMINOLOGIES
In such competitive and creative environment, new terms
and phrases have born to address the current requirement and
demands of the industry. Recently, Germany has announced
the Industry 4.0 methodology as the fourth industrial
revolution. Three past revolutions are "First mechanical loom"
in 1784, "First assembly line" in 1870, "First programmable
logic controller (PLC)" in 1969 respectively. Based on the
industry 4.0 terminology, intelligent data analysis and
interconnected systems are combined together to generate a
brand new aspect in factory transformation and production
management. Table 1 represents the differences between a
today's factory and an Industry 4.0 factory. In current industry
environment, providing high-end quality service or product
with the least cost is the key to success and industrial factories
are trying to achieve as much performance as possible to
increase their profit as well as their reputation. In this way,
various data sources are available to provide worthwhile
information about different aspects of the factory. In this stage,
the utilization of data for understanding the current condition
and detect faults and failures is an important topic to research.
e. g. in production, there are various commercial tools
available to provide OEE (Overall Equipment Effectiveness)
information to factory management in order to highlight root
cause of problems and possible faults in the system. In
contrast, in an Industry 4.0 factory, in addition to condition
monitoring and fault diagnosis, components and systems are
able to gain self-awareness and self-predictiveness, which will
provide management with more insight on the status of the
factory. Furthermore, peer-to-peer comparison and fusion of
health information from various components provides a precise
health prediction in component and system levels and enforce
factory management to trigger required maintenance at the best
possible time to reach just-in time maintenance and gain near
zero downtime. In addition to Industry 4.0, Cyber-physical
system (CPS) is a phrase representing the integrated
computational and physical capabilities such as sensing,
communication and actuation to physical world[3], which is
addressed by American government since 2007 as a new
developments strategy[4],[5].

Proceeding of Int. Conference on Industrial Informatics (INDIN) 2014
Applications of CPS include but not limited to
manufacturing, Secure Control, medical devices,
Environmental Control, aviation, advanced automotive
systems, process control, energy control, traffic control and
safety, smart structures and so on[6]. Having in mind that CPS
is in its immature stage and it covers a broad range of scientific
era, interactive collaborations between industry and academia,
especially long-term collaborations such as the tasks that have
been done in Advanced Research and Technology for
Embedded Intelligence Systems (ARTEMIS) in Europe or
Center for Intelligent Maintenance Systems (IMS) in United
States will help in identification of challenges and significantly
accelerate the progress of CPS[4], [7].
In the implementation of advanced terms such as Industry
4.0 and Cyber Physical Systems, sole presence of connectivity
between machines and using sensors is not useful. To leverage
these advanced technologies, correct information has to be
present at the right time for the right purpose. In this situation,
a 6C system that is consist of Connection (sensor and
networks), Cloud (data on demand), Cyber (model & memory),
Content (meaning and correlation), Community (sharing &
collaboration), and Customization (personalization and value)
[7] can enhance the information system. In this scenario and in
order to provide useful insight to the factory management and
gain correct content, data has to be processed with advanced
tools to generate relative information. Considering the presence
of visible and invisible issues in an industrial factory, the
information generation algorithm has to capable of detecting
and addressing invisible issues such as machine degradation,
component wear, etc in the factory floor.
III. PROGNOSTICS AND HEALTH MANAGEMENT
Prognostics and health management (PHM) is an enrich
research area that mostly deals with component wear and
degradation. Factory wide transparency is one of the most
important targets of PHM because in contrast with visible
issues, invisible issues might happen due machine degradation,
component wear and etc. while operators and factory managers
are not aware of them. Indentifying those issues can lead to
serious downtimes in the factory floor and PHM as an ever-
growing pioneer research domain provides useful approaches
to detect those invisible problems and avoid unplanned
downtimes to overwhelm uncertainties in the systems.
Remaining useful life prediction, fault diagnosis and fault
detection are among the useful tools provided by PHM
algorithms for bringing transparency in every aspect of future
industry by analyzing sensory and system level data. In Past
few years, different aspects of industry as well as different
machines or components have been targeted by researchers for
developing reliable PHM solutions. Aircraft engines [8],
industrial robots [9], machine tools [10], electrical motors[11],
wind turbine[12], batteries [13], gearboxes [14], bearings [15],
pumps [16] and etc are only few examples of a broad range of
assets.
IV. INTEGRATED SYSTEMS
In previous sections, we introduced Industry 4.0 and Cyber-
Physical Systems as two leading infrastructures for managing
data and leaning toward more efficient production in current
industry. In addition, we presented prognostics and health
management methods as powerful and applicable approaches
for detecting invisible faults and failures among factories fleet.
In current section, advantages of implementing PHM
algorithms in Industry 4.0 and CPS application will be
presented and the next section introduces a methodology for
implementing cyber-physical systems in industry 4.0
applications.
Today, most PHM methods have only access to target asset
data. Asset data might contain hundreds of sensory and system
data readings and provide good insight and prognostics results
for the current asset. But, there are much more sources of
information that are not considered in today's implementation
of PHM methods. Peer-to-peer evaluation through fleet of
assets, life-cycle historical data from identical assets and
system configuration are few examples of the huge portion of
data that is ignored by now.. For example, in PHM, various
failure signatures are required for accurate fault diagnosis of
assets and improve the reliability of maintenance scheme.
Table 1: Comparison of today’s factory and an Industry 4.0 factory

Proceeding of Int. Conference on Industrial Informatics (INDIN) 2014
Acquiring every possible failure signature from one instance of
assets is not likely to happen but in an interconnected system
which manages hundreds of similar assets among the fleet, the
occurrence likelihood of various failures much higher.
Therefore, the PHM analytical engine can capture those failure
signatures from various assets to refine its failure detection and
estimation capabilities. This huge improvement is possible by
utilizing the concept of cyber-physical systems for PHM
applications in which a cyber twin model (avatar) of real
machine is created and operated in cloud
platform parallel to actual asset. Leveraging the Integrated
knowledge available in cyber physical model, health condition
of assets can be accurately simulated and appropriately
presented to dedicated users upon their demand without
geographical limitations. Furthermore, the interconnectivity of
fleet of identical assets to their cyber-physical model not only
provides the opportunity of peer-to-peer evaluation and
prognostic library accumulation but also enables PHM
algorithms to have access to various life stages of different
assets as well as test stage data.
V. METHODOLOGY FOR DESIGNING CPS BASED
INDUSTRY 4.0 SYSTEMS
Knowing the capabilities of cyber-physical systems, a
promising methodology for designing Industry 4.0 applications
based on Cyber-Physical Systems can be developed. As it was
discussed in previous section, interconnectivity provides access
to vast amount of data. But, sole availability of data does not
create a significant advantage. Therefore, an adaptive yet
powerful methodology is required to manage, categorize and
process data for further analysis by PHM algorithms. This
method has to be broad enough to truly leverage all the
advantages of cyber-physical systems.
In this article, we propose “Time Machine Methodology for
Cyber-Physical Systems” which is in charge of perfectly
organize available data in Big Data environment to be prepared
for usage in PHM algorithms. Every single component of the
fleet will have a representative Time Machine record in the
cyber space. This cyber delegate extracts worthwhile
information from the pool of available data and normalizes it
for further analysis. Extracted information include but not
limited to implementation history, stress and load, operation
parameters, system configurations and maintenance records.
Once the actual component is failed, it will be removed from
the parent machine and will no longer exist to participate in
analytics. But its cyber twin (Time machine record) will
remain without any time constraints. Unlimited existence of
cyber twins results in continuous accumulation of Time
Machine records and consequently gathering various operation
parameters from broad range of identical components.
Normalization of parameters, in which further research efforts
have to be conducted, ensures the comparability of time
machine records with each other for identical components.
Additionally, time machine records obey the hierarchical
relation of actual components as well and every cyber twin has
access to records of its predecessor and ancestor components.
Such information rich environment brings significant
robustness to PHM algorithms for continuous and accurate
predicting and monitoring of the factory. Ultimately, this
methodology brings ultimate implementation of cyber physical
system into action for designing an Industry 4.0 factory.
Fig. 1. The general framework of the application of PHM algorithms in detecting invisible issues in industry using Watchdog Agent® tool

Proceeding of Int. Conference on Industrial Informatics (INDIN) 2014
VI. CASE STUDIES
A. Industrial Robot
In this section we consider the development of a cyber
physical system for health monitoring of industrial robots. Fig.
2. shows the schematic view of the cyber-physical model for
current case study on industrial robots. This study was focused
on preparing a predictive health monitoring solution for a fleet
of 30 industrial robots in production line. Due to variety of line
speeds, a more complicated multi-regime approach had to be
undertaken for establishing a robust prognostics and health
management algorithm as the core analytics of cyber physical
system by using torque and speed data. Due to its non-invasive
nature torque monitoring is a popular fault detection method
for monitoring health condition of industrial robots and
therefore, most of the research efforts in this area have been
focused on this parameter. In addition, nonlinear relation
between operating speed and torque cast a challenge on PHM
algorithms to correctly determine the health state of the robot.
In addition to condition data (torque and speed) the cyber
physical model obtains various configuration parameters such
as gear ratio, load ratio, pressure calibration, type of tooling for
robot servo guns and assigned products to specific robots from
production line.
Fig. 2. Cyber-Physical Model for Industrial Robot Monitoring
Fig. 3. Methodology for Designing CPS based Industry 4.0 Systems

Proceeding of Int. Conference on Industrial Informatics (INDIN) 2014
These configuration parameters help the core of the model to
standardize and adaptively cluster the operation data for more
accurate processing. Ultimately, the entire analytical engine of
this study was set and established on the cloud where the whole
dataset (condition and configuration data) were stored on the
cloud storage and health monitoring algorithm used the data
stored on the cloud (from entire fleet) to calculate the health
condition of each individual instance. The outcome of PHM
algorithms were presented to users as enriched infographics
through a web-based user interface
B. Virtual Battery
The battery pack is one of the most significant components
of an Electric or Hybrid Vehicle. The uncertainty in the driving
range, the batteries reliability and service life, and battery
safety concerns, are all challenges that must be overcome in
order to achieve more widespread adoption of electric vehicles.
In order to meet the challenges and rising demand in hybrid
and electric vehicles in recent years, the importance of batteries
has become more and more important[17]. Therefore, fully
understanding of the dynamic performance of batteries under
various conditions is of significant importance. However,
batteries operation condition includes several stress factors
such as environment temperature, humidity, driving style,
charging level, discharge rate and road conditions. To achieve
this battery understanding a battery model is needed which
should provide capabilities for health evaluation and failure
prediction, through simulation of cell performance under
different conditions. Additionally, using such a battery model,
the functionality of a manufactured battery can be fed back to
design and manufacturing suppliers to recognize the potential
impact of design and enhancement of manufacturing process
on battery performance[18].
An overview of the electric vehicle and battery health
management and prognostic platform is illustrated in fig. 4,
which includes algorithms for state of charge and state of
health estimation and driving behavior classification.
Presently, most of the battery models focus on individual
battery cells, or on battery package not including the detailed
dynamics within the battery. However, all components in a
battery are associated with each other. The interactions among
all the cells, conductors, BMS, and environment temperature
will play a significant role in battery performance, and deserve
an integrated study. Moreover, manufacturing practice strongly
impact on battery application conditions & user behavior.
Therefore a simulated framework is needed which is
capable of execution the functions discussed above by
integrating various models to emulate the multi-regime
changes in battery parameters and inputs and investigate their
impacts on battery functionality .
A "Virtual Battery" is a cyber-physical model resulting
from merging the new battery technology with intelligent
enabling tools to provide following solutions:
1. Enabling prognostics tools to transform data to health
information regarding the health, reliability and operational
readiness of the battery system.
2. Adequate visualization system to deliver the right
information to the right people. Some information that requires
the immediate intervention of the driver may be displayed for
the actual user (driver) on the dashboard, while other forms of
information (including more detailed diagnostic information)
would be necessary for a maintenance personnel or logistics
center and used to schedule maintenance or replacements.
3. Enabling tether-free communication to capture this
information in real-time from the vehicle and deliver it to a
central logistics center that will advise the driver about his
battery condition and could use GPS trip information to direct
the driver to the nearest service center or rapid charging center
on his trip.
4. Simulating and predicting battery behavior such as
SoC (State of Charge) and SoH (State of Health) in other
operation conditions based on external parameters(
environment , driving style, etc ) and internal parameters (
battery type, and battery age) without take a long time to test a
battery in that condition.
5. Help the designers to find the flaws or design issues
for a better hybrid cycle management in parallel with a better
estimation of the selected pack’s lifetime expectancy in the
specific application.
6. Help the battery manufacturers to locate the faulty
packs and study them to refine their design, manufacturing
process or raw material selection.
7. Integration of the wireless solutions into the
prognostic tool could also open an opportunity for better
monitoring of the vehicle during standard or extended warranty
period. The faulty system could be recalled for repair long
before the usual inspection periods. This in turn can reduce the
costs of warranty that could rise from such occasions. There is
also a possibility to integrate this system into engine
management sensors. This will enable the designers to study
the battery aging process in different climatic conditions that
the vehicle operates in. This over-time data acquisition could
be an easy and cheap way to collect field data. These data
could be used to design the next generation systems in a better
way.
VII. CONCLUSION
In this article we introduced recent advances in industrial
informatics with respect to Big Data environment, Cyber-
Physical Systems and Industry 4.0. In addition, the importance
of intelligent prognostics and health management in industry
for retaining production and service excellence has been
discussed. Furthermore, the fusion of cyber-physical systems
and PHM algorithms in big data environment and the
advantage of using interconnected systems has been presented
along with two industrial application of cyber-physical
systems.
Current industrial evolution is guiding industry toward
maximum leverage from benefits of interconnected systems in
big data environment where companies with more futuristic
vision that establish new methodologies in their culture will
have the opportunity of being significantly successful and
profitable in recent future.

Proceeding of Int. Conference on Industrial Informatics (INDIN) 2014
REFERENCES
[1] L. Da Xu, “Enterprise systems: state-of-the-art and future trends,”
Industrial Informatics, 2011.
[2] E. Dumbill, “Big Data is Rocket Fuel,” Big Data, 2013.
[3] “Adaptive calibration for fusion-based cyber-physical systems,” vol. 11,
no. 4, pp. 1–25, Dec. 2012.
[4] “Cyber-physical systems,” IEEE Control Systems Society, 2011.
[5] J. Shi, J. Wan, H. Yan, and H. Suo, “A survey of Cyber-Physical
Systems,” Signal Processing (WCSP 2011), pp. 1–6, 2011.
[6] “Cyber Physical Systems: the need for new models and design
paradigms,” Carnegie Mellon University.
[7] J. Lee, E. Lapira, B. Bagheri, and H. Kao, “Recent advances and trends
in predictive manufacturing systems in big data environment,” Manufacturing
Letters, 2013.
[8] T. Wang, J. Yu, D. Siegel, and J. Lee, “A similarity-based prognostics
approach for Remaining Useful Life estimation of engineered systems,” 2008
International Conference on Prognostics and Health Management (PHM), pp.
1–6, 2008.
[9] E. R. Lapira, “Fault detection in a network of similar machines using
clustering approach,” 2012.
[10] L. Liao and J. Lee, “Design of a reconfigurable prognostics platform for
machine tools,” Expert systems with applications, vol. 37, no. 1, pp. 240–252,
2010.
[11] B. Bagheri, H. Ahmadi, and R. Labbafi, “Application of data mining and
feature extraction on intelligent fault diagnosis by Artificial Neural Network
and k-nearest neighbor,” 2010 XIX International Conference on Electrical
Machines (ICEM), pp. 1–7, 2010.
[12] E. Lapira, D. Brisset, H. D. Ardakani, and D. Siegel, “Wind turbine
performance assessment using multi-regime modeling approach,” Renewable
Energy, 2012.
[13] M. AbuAli, E. R. Lapira, W. Zhao, D. Siegel, M. Rezvani, and J. Lee,
“Systematic Design of Prognostics and Health Management Solutions for
Energy Applications,” in Engineering Asset Management 2011, no. 22, J. Lee,
J. Ni, J. Sarangapani, and J. Mathew, Eds. London: Springer London, 2013,
pp. 243–250.
[14] B. Bagheri, H. Ahmadi, and R. Labbafi, “Implementing discrete wavelet
transform and artificial neural networks for acoustic condition monitoring of
gearbox,” Elixir Mech Engg, 2011.
[15] O. D. Snell and I. Nairne, “Acoustic bearing monitoring—the future
RCM 2008,” Railway Condition Monitoring, 2008.
[16] A. Soylemezoglu and S. Jagannathan, “Mahalanobis-Taguchi System as
a Multi-Sensor Based Decision Making Prognostics Tool for Centrifugal
Pump Failures,” Reliability, IEEE Transactions on, vol.60, no.4, pp.864,878,
Dec. 2011.
[17] F. Ju, J. Wang, J. Li, and G. Xiao, “Virtual battery: A battery simulation
framework for electric vehicles,” Automation Science and Engineering, IEEE
Transactions on , vol.10, no.1, pp.5,15, Jan. 2013.
[18] Y. He, W. Liu, and B. J. Koch, “Battery algorithm verification and
development using hardware-in-the-loop testing,” Journal of Power Sources,
2010.
Raw Data
Voltage
Current
Temperature
Speed
GPS data
EV Weight
Historical
Data
Intelligent
Algorit hm
Cyber-Physical Battery Platform
Driver
SoC Estimatio n
Driving Behavior
Classification
EV Manufacturer
EV Range
EV Productivit y
EV Safety
Battery Supplier
Material Development
Battery Safety
Information
Real
Condition
Voltage
Current
Temperature
Speed
GPS data
EV Weight
Virtual
Condition
Simulated Data
Predicting Results
Fig. 4. Cyber Physical Battery Platform