![Global deployment of RERs [9].](https://www.researchgate.net/profile/Muhammad-Fahad-Zia/publication/324835813/figure/fig1/AS:624704528347136@1525952574365/Global-deployment-of-RERs-9_Q320.jpg)
![MG market status [9].](https://www.researchgate.net/profile/Muhammad-Fahad-Zia/publication/324835813/figure/fig2/AS:624704528322561@1525952574463/MG-market-status-9_Q320.jpg)
Content uploaded by Muhammad Fahad Zia
Author content
All content in this area was uploaded by Muhammad Fahad Zia on May 21, 2018
Content may be subject to copyright.
MICROGRIDS ENERGY MANAGEMENT SYSTEMS: A CRITICAL REVIEW ON
METHODS, SOLUTIONS, AND PROSPECTS
Muhammad Fahad Ziaa, Elhoussin Elbouchikhib, Mohamed Benbouzida,c
aUniversity of Brest, UMR CNRS 6027 IRDL, 29238 Brest, France
bISEN Yncr´ea Ouest, UMR CNRS 6027 IRDL, 29200 Brest, France
cShanghai Maritime University, 201306 Shanghai, China
Abstract
Renewable energy resources are currently being deployed on a large scale to meet the requirements of increased energy demand,
mitigate the environmental pollutants, and achieve socio-economic benefits for sustainable development. The integration of such
distributed energy sources into utility grid paves the way for microgrids. The microgrid concept is introduced to have a self-
sustained system consisting of distributed energy resources that can operate in an islanded mode during grid failures. In microgrid,
an energy management system is essential for optimal use of these distributed energy resources in intelligent, secure, reliable, and
coordinated ways. Therefore, this review paper presents a comparative and critical analysis on decision making strategies and
their solution methods for microgrid energy management systems. To manage the volatility and intermittency of renewable energy
resources and load demand, various uncertainty quantification methods are summarized. A comparative analysis on communication
technologies is also discussed for cost-effective implementation of microgrid energy management systems. Finally, insights into
future directions and real world applications are provided.
Keywords: Microgrid, renewable energy resources, communication technologies, energy management system,
optimization.
Nomenclature
CG Conventional generator
DER Distributed energy resource
DR Demand response
EMS Energy management system
ESS Energy storage system
EV Electric vehicle
GA Genetic algorithm
GHG Greenhouse gas
LC Local controller
LP Linear programming
MAS Multi-agent system
MG Microgrid
MGCC Microgrid central controller
MILP Mixed integer linear programming
MPC Model predictive control
NN Neural network
PEI Power electronic interface
PSO Particle swarm optimization
RER Renewable energy resource
Email addresses: muhammadfahad.zia@univ-brest.fr (Muhammad
Fahad Zia), elbouchikhi@isen-bretagne.fr (Elhoussin Elbouchikhi),
mohamed.benbouzid@univ-brest.fr (Mohamed Benbouzid)
1. Introduction
The exponential increase in global energy demand is the
main cause of rapid depletion of fossil fuels and increased green-
house gas (GHG) emissions of conventional generators (CGs).
To overcome these problems, the world has taken initiatives to
deploy renewable energy resources (RERs) on a large scale, in
order of GW, since a decade [1, 2]. The RERs, such as solar,
wind, biomass, hydro, and tidal power are one of the most im-
portant sources in providing clean energy and mitigating GHG
emissions for sustainable development [3]. United Nations Sus-
tainable Development and Paris Climate Agreement goals also
promote installation of RERs. In 2016, the global deployment
of RERs, excluding hydro power, reached 921 GW due to the
increased awareness of climate change, with China being at the
top spot in deployment of RERs followed by the USA, and Ger-
many, as shown in Figure 1.
RERs, micro CGs, and energy storage systems (ESSs) are
often described as distributed energy resources (DERs) in the
literature [4]. DERs are on-site generation sources in distri-
bution system. Hence, no transmission equipment is required
for power transfer to load ends. In DERs, the RERs, partic-
ularly solar and wind energy, are volatile and intermittent en-
ergy sources. Therefore, ESSs and micro CGs are needed to
overcome these uncertainties. The integration of DERs into
distribution network requires the optimal sizing, control, and
scheduling of these energy resources. A microgrid (MG) em-
bodies these issues by integrating DERs into power grid, to-
Preprint submitted to Applied Energy May 10, 2018
gether with an ability to operate in an islanded mode during
main grid failure [5]. Hence, It helps in achieving objectives
of efficient transformation of the passive network into an active
one, bidirectional and controlled power flow management, re-
liable and continuous supply, power quality enhancement, and
clean environment.
MG is defined as a low-voltage distribution network of in-
terconnected DERs, controllable loads, and critical loads. It
can operate in either grid-connected or islanded mode subject
to operational characteristics of the main grid [6, 7]. To have
a flexible operation of MG, power electronic interfaces (PEIs)
and controls are integrated with DERs for power quality moni-
toring, continuous, and reliable power supply [8].
MGs offer several advantages such as reduction in GHG
emissions, reactive power support for voltage profile improve-
ment, decentralization of energy supply, heat load integration
for cogeneration, ancillary services and demand response (DR)
[8]. It also reduces line losses and outages in transmission and
distribution system. The MG deployment market status, all over
the world, is presented in Figure 2. Autonomous basic MGs are
defined as microgrids that cannot meet the load demand for 24
hours a day. However, autonomous full MGs have provisions
for supplying energy to load end for all the day. Autonomous
basic MGs are already deployed in all over the world except in
Middle East, while autonomous full and interconnected com-
munity-based MGs are still in emerging and piloting phases of
deployment.
MGs have some limitations such as high investment cost of
RERs, optimal use of energy sources, control issues and lack of
system protection and regulatory standards, and customer pri-
vacy. Due to high deployment of inherently intermittent RERs
and increased integration of probabilistic controllable loads into
MGs, researchers have focused on solving its energy manage-
ment problems.
Energy management system (EMS) of an MG encompasses
both supply and demand side management, while satisfying
system constraints, to realize an economical, sustainable, and
reliable operation of MG [10]. EMS provides many benefits
from generation dispatch to energy savings, reactive power sup-
port to frequency regulation, reliability to loss cost-reduction,
energy balance to reduced GHG emissions, and customer par-
Figure 1: Global deployment of RERs [9].
Figure 2: MG market status [9].
ticipation to customer privacy.
Several review papers addressed different aspects of MGs,
as a survey on experimental MG systems installed in Europe,
North America, and Asia regions [17], protection and control
schemes for MG [18], and reactive power compensation tech-
niques in MG [19]. They also addressed control methods for in-
verter-based MGs [20], droop control techniques [21], and con-
trol strategies for voltage and frequency regulation of MG [22].
They examined control strategies for DERs in MG [23], mod-
eling, design, planning and architectures of hybrid renewable
MGs [24], and a survey on AC and DC MGs [25]. The litera-
ture also provided reviews on homeostatic control-based energy
efficient micro-generation systems [26], MG uncertainty quan-
tification methods [27], and the survey on energy efficiency in
buildings and MGs using network technologies [28]. Review
papers related to energy management of MG are summarized
by Table 1. Unlike these review papers, this paper presents
a comprehensive and critical analysis of the MG EMSs. MG
EMSs based on solution approaches like robust optimization,
hierarchical control, homeostatic control, chance constrained
programming, mesh adaptive direct search, and meta-heuristic
approaches such as gravitational search, bacterial foraging, and
many others have not been previously addressed. Moreover,
the uncertainty quantification methods used in each MG EMSs
are also highlighted. Furthermore, deployment cost, coverage
range, and data rate-based graphical comparison on communi-
cation technologies selection is also discussed. Such compari-
son is necessary for the achievement of a secure, economical,
and reliable communication network among MG components.
It is also important for a smooth operation of centralized and,
particularly, decentralized architecture of MG. Finally, insights
into future directions and real world applications are also pro-
vided.
This review paper presents a comparative analysis on the
proposed MG EMSs. Section 2 provides an overview on MG
architectures and MG classification. Communication technolo-
gies used for effective coordination among MG components are
discussed in section 3. Section 4 provides a critical review
on different MG EMS strategies based on various solution ap-
proaches used by authors and their main limitations. Section
5 discusses real world applications of MG EMSs followed by
insights into future directions.
2. Microgrid Architecture
An MG is composed of different DERs, responsive loads,
and critical loads, as shown in Figure 3. The MG is connected
2
Table 1: Existing reviews related to energy management of MG.
Ref. Details
[11] Modeling of RERs and ESSs are briefly explained. This review also discusses meta-heuristic optimization methods and software tools used for energy management and
control of hybrid RERs, sizing objectives, ESS management, power quality, and energy dispatch related problems.
[12] Authors conducted a comprehensive review on energy management in MGs. The review topics are optimization objectives, constraints, algorithm types, and software
tools.
[13] Authors conducted a comprehensive review on centralized EMSs of MG based on objectives such as power management, economic dispatch, and unit commitment
using solution approaches of mixed integer linear and nonlinear programming methods, genetic algorithm, particle swarm algorithm, rule-based system, and artificial
intelligent methods. Multi-agent system (MAS) is explained extensively for optimization of decentralized EMS of MG.
[14] Potential of multi-agent systems is discussed in detail in the context of energy management, operation, security and stability of MG system.
[15] Energy management strategies in the context of stand-alone and grid-connected hybrid RERs systems are reviewed. Energy management strategies based on linear
programming and intelligent techniques are discussed.
[16] Authors presented a comprehensive review on MG planning. This review points out different computational optimization techniques applied to scheduling, reliability,
environmental, sizing, and siting problems.
LC
Renewable energy resources
Conven�al genera�on sources
Energy storage systems
Smart homes
Smart buidlings
Electric Vehicles
Main Grid
LC
LC
LC
LC
LC
LC
Weather
forecasts Energy
market
Microgrid
Central
Controller
Informa�on and
communica�on flow
Energy flow
Figure 3: MG architecture.
to the main grid through a point of common coupling (PCC)
[55]. In both grid-connected and islanded modes, each DER
is connected with PEI to achieve control, metering, and pro-
tection objectives together with an ability of a plug and play
feature. During grid-connected mode, an MG reaps advantages
of power trading with the main grid. However, in case of dis-
turbances or failure in main grid, MG shifts its operation to is-
landed mode to ensure system stability. In this mode, it provide
continuous supply to critical loads by efficient integrated oper-
ation of DERs, DR, and load shedding (LS). The entire MG op-
eration is controlled and coordinated by MG central controller
(MGCC) and local controllers (LCs) [56]. The effective man-
agement and coordination of DERs in MG results into improved
system performance and sustainable development [57].
Due to the increased awareness of climate change, socio-
economic development, and the need to mitigate GHG emis-
sions, MGs mainly consist of sustainable energy systems, as
renewable energy systems and energy efficient systems that use
local heat waste [58]. Optimization of these energy systems is
achieved by MG EMS that solves decision making strategies.
These strategies consider increased system energy efficiency,
increased reliability, reduced energy consumption, decreased
3
Table 2: Sustainable energy systems in MG.
Ref. Solar Wind Fuel cell Combined heat
and power
Biomass Hydro Tidal
[29],[30],[31],[32] X X
[33] X X
[34] X X X
[35],[36],[37] X X
[38] X X X
[39] X X
[40],[41],[42] X X X
[43] X X X
[44] X X X
[45] X X X
[46] X X X X
[47],[48],[49],[50] X X X
[51] X X X
[52] X X
[53] X X X
[54] X X X
Microgrid Classifica�on
Opera�on
mode
Power
type
Supervisory
Control Applica�onPhase
AC DC CentralizedDecentralized Islanded Grid-
connected
Single
Phase
Three
Phase
Residen�al
or
commercial
or
Industrial
U�lity
or
muncipility
or
military
Figure 4: MG classification.
operational cost of DERs, reduced system losses, and mitiga-
tion of GHG emissions for sustainable development. Table 2
presents few examples of sustainable energy systems that are
used in the literature for energy management operation of MG.
Apart from mode of operation, MGs are also classified with
respect to power type, supervisory control, supply phase, and
application as shown by Figure 4.
3. Microgrid Communication
Dispersed generation of DERs and active integration of DR
require a communication infrastructure to share information with
each other and optimize their operation locally [59, 60]. There-
fore, an efficient data communication system is needed for con-
tinuous, fast, reliable, and accurate transfer of information among
sensors, LCs and MGCC without any disturbances and discon-
nections. However, investment cost of such data communica-
tion systems can be much high, which depends upon the num-
ber of repeaters required to improve the quality of transmitted
signals, while covering specific geographical area. Hence, it is
vital to reduce installation cost, while maintaining reliable op-
eration, by selecting suitable data communication technology
for short and long distance applications [61, 62].
In the literature, several wired and wireless communica-
tion technologies have been proposed for effective communi-
cation among different MG components. The selection of these
communication technologies depends on the data rate, coverage
area, quality of service, reliability, latency, and power consump-
tion [63]. An overview of different communication technolo-
gies that can be used for MG operations, is shown in Figure 5.
Among these communication technologies, wired technologies,
such as DSL, PLC, and fiber optics, have higher data transmis-
sion rate and reliability but at expense of high installation cost.
On the contrary, wireless technologies, such as Zigbee, Z-wave,
GSM, and wifi, etc., can be easily deployed with lesser installa-
tion cost, hence being better candidates for remote areas. How-
ever, they have low data transmission rate and signal interfer-
ence problems. To conclude, with recent advancements in MGs
installation, more sensors, meters, and LCs are needed to be in-
tegrated, monitored, and controlled regularly. Therefore, wire-
less technologies are, overall, better candidates as compared to
wired ones due to their low deployment cost.
4. Microgrid Energy Management System
The International Electrotechnical Commission in the stan-
dard IEC 61970, related to EMS application program interface
in power systems management, defines an EMS as “a computer
system comprising a software platform providing basic support
services and a set of applications providing the functionality
needed for the effective operation of electrical generation and
transmission facilities so as to assure adequate security of en-
ergy supply at minimum cost” [64]. An MG EMS, also hav-
ing these same features, usually consists of modules to perform
decision making strategies. Modules of DERs/load forecasting,
Human Machine Interfaces (HMI), and supervisory, control and
data acquisition (SCADA) among others ensure the efficient im-
plementation of EMS decision making strategies by sending op-
timal decisions to each generation, storage, and load units [65].
4
Data rate (bps)
100 K 1 M 10 M
100
1 K
10 K
100 K
>1 M
NB-PLC
BB-PLC
PON
DSL
Zigbee Wifi
100 M >1 G
Satellite
2G
3G
4G
Bluetooth
Wimax
10
Coaxial Cable
Lowest Deployment Cost Highest
10 K
1 K
Coverage Area (m)
NB-PLC : Narrow band power line communica�on
Z-wave
BB-PLC : Broad band power line communica�on
PON : Passive op�cal network
DSL : Digital Subscriber line
Figure 5: Communication technologies for MG operation.
An MG EMS performs variety of functions as monitoring, ana-
lyzing, and forecasting of power generation of DERs, load con-
sumption, energy market prices, ancillary market prices, and
meteorological factors as given by Figure 6. These functions
help EMS in optimizing MG operation, while satisfying the
technical constraints.
The supervisory control architecture of MG EMS can be
divided into two types, namely, centralized and decentralized
EMSs. In centralized EMS, the central controller accumulates
all the information such as power generation of DERs, cost-
function, meteorological data, and energy consumption pattern
of each consumer, etc. Then centralized EMS determines the
optimal energy scheduling of MG and sends these decisions
to all LCs. However, in decentralized EMS architecture, the
MGCC sends and receives all the information to LCs in real-
time. Each LC proposes a current and future demand or gener-
ation request to the MGCC. The MGCC determines the optimal
scheduling and sends it back to the LC. The latter may dis-
agree with the current operation and continue to bargain until
the global and local objectives are achieved. With the integra-
tion of RERs, ESSs, EVs, and DR, the MG EMS strategies have
been diversified from economic dispatch and unit commitment.
Data Monitoring
Data Analy�cs
Forecas�ng
Op�miza�on Energy Market
Ancillary Market
Real �me control
RERs
CGs
ESSs
DR
EVs
Grid
Weather forecasts
Figure 6: MG EMS functions.
The other strategies are scheduling of DERs and loads, mini-
mization of system losses and outages, control of intermittency
and volatility of RERs, and realization of economical, sustain-
able, and reliable operation of MG. These MG EMS strategies
are shown in Figure 7.
Many researchers have solved these energy management
strategies using various solution approaches to achieve the opti-
mal and efficient operation of MG. An extensive critical review
of these strategies and solution approaches is described in the
following subsections.
4.1. EMS based on Classical Methods
4.1.1. EMS based on Linear and Nonlinear Programming Meth-
ods
Sukumar et al. [35] proposed power sharing, continuous
run, and on/off-based mixed mode MG EMS. The power shar-
ing mode considers the power trading with the main grid, while
fuel cell is bound to remain operational in continuous run mode.
Both of these modes are solved by a linear programming (LP)
optimization method. However, on/offmode is solved by a
mixed integer linear programming (MILP) solution approach
that optimizes the operation of MG with respect to on/offcon-
nection status of main grid, fuel cell, and ESS. The sizing of
ESS is also computed considering MG operational requirements.
A techno-economic MILP-based optimization model is propos-
ed in [29] to compute energy consumption scheduling in MG
EMS. The authors also presented advantages of DR program in
handling intermittency and volatility effects of RERs, improv-
ing load factor, and reducing peak consumption. The capital,
replacement, operational and maintenance (O&M), and resid-
ual costs-based MILP net present cost-optimization model is
solved by GAMS software. HOMER simulation software is
used for sizing optimization of MG.
Anglani et al. [66] proposed optimal EMS for remote mili-
tary MG. The piecewise linear fuel consumption model of diesel
generator is used as an objective function. The Rainflow count-
ing method is also used to determine the battery sizing based
on its lifecycle and depth of discharge to tradeoffbetween op-
erational and capital costs of MG. The special order sets 1 and
2 based branch and bound solver tool solve the proposed semi-
continuous optimization model for two scenarios. In scenario
1, only one generator operates, while in scenario 2, both gen-
erators are operating but only one will be used at a time. The
effectiveness of the proposed optimal EMS model is also ex-
perimentally validated. Comodi et al. [67] presented a MILP
energy trade profit model-based EMS for residential MG. Ra-
dial basis neural network (NN) method is used to forecast the
power output of photovoltaic (PV) and solar thermal power sys-
tems. The profitable integration of thermal storage is realized
by heat load management. However, the batteries are concluded
to be economically unfeasible for integration in residential mar-
ket due to their high investment and replacement costs.
For optimal energy management of residential MG, an op-
erating cost-minimization model is proposed in [68]. It in-
corporates energy trading cost, penalty cost on adjustable load
shedding, EVs batteries wearing cost, the range anxiety term
5
Objec�ves
Constraints
Figure 7: Energy management strategies of MG.
for electric vehicles (EVs). Three types of load are considered
namely: critical, adjustable, and shiftable loads. The range anx-
iety of EVs is defined as the fear factor of running out of energy
before reaching destination. Risk level-based three scenarios in
range anxiety term are studied to analyze the tradeoffbetween
operational cost of MG and average state of charge (SOC) of
EVs battery. Shen et al. [69] proposed an optimal revenue
maximization model of MG EMS together with main grid peak
shaving application by introducing demand responsive loads.
The authors assumed that the load demand of MG will always
remain more than its generation supply. Two case studies of one
bus and IEEE 14-bus MG systems are used to analyze the per-
formance and effectiveness of the proposed MILP optimization
model in CPLEX software. In [47], Tenfen and Finardi pro-
posed an optimal energy management strategy to minimize the
operational cost of MG together with incorporation of curtail-
able and shiftable loads in DR. The objective function includes
the O&M costs, start-up and shutdown cost, energy trading cost
with main grid, and load shedding cost.
Vergara et al. [70] proposed security constrained EMS for
three-phase residential MG. The nonlinear optimization model
is developed to minimize the operational cost of MG, while pe-
nalizing load shedding. The system outages are included as a
constraint to ensure reliability of MG. The developed nonlinear
model is converted into MILP model by linearizing the objec-
tive function, active and reactive power of generation sources,
ESS model, and bus phase angles. Piecewise approximation,
Taylor series, estimated operation points, and introduction of
auxiliary variables-based methods are used for it. The perfor-
mance and efficiency of MILP approximate model is compared
with the three-phase nonlinear optimal power flow problem that
results in an energy supply error below 2% with lesser com-
putational time. Luna et al. [71] presented a real-time on-
line MG EMS to minimize the operational and load shedding
costs. Three case studies of perfect forecast, imperfect forecast,
and accurate information are used in analyzing the developed
model. A comparative analysis is performed among these three
case studies in respect of savings and computational time.
A mixed integer nonlinear energy management model of an
isolated three-phase unbalanced MG is proposed in [72]. It
minimizes the fuel, start-up and shutdown costs of CGs, and
penalty cost on reactive power requirements. The mathematical
models and ABCD parameters of CG, ESS, transmission line,
and transformer are formulated. The developed MINLP model
is divided into two sequential stages of MILP-based unit com-
mitment and nonlinear programming-based optimal power flow
models. This decomposition solves the problems of intractabil-
ity and non-convergence. Helal et al. [73] have proposed a
mixed integer nonlinear EMS model for optimal operation of an
islanded MG considering DR. The developed model includes
operating cost of CGs as an objective function. Furthermore,
droop controlled active and reactive power dispatch of AC side
CGs, and operation of water desalination units are also included
as a constraint in the proposed model.
Tsikalakis and Hatziargyriou [74] developed centralized ar-
chitecture for energy management of a grid-connected MG. Two
market policies are proposed to determine the price bids for
MG participation in energy market. The objective of first pol-
6
icy is operational cost-minimization of MG, while second pol-
icy aims to maximize its profit considering energy transactions
with main grid. Both of these optimal policies are solved using
sequential quadratic programming method. Panwar et al. [37]
introduced strategic EMS of a grid-connected MG, constrained
by an operation window of transformer nominal operation and
voltage security. The developed model minimizes MG oper-
ational cost using modified gradient descent solution method.
The forward backward sweep algorithm determines power flow
solution of MG. Three scenarios are considered in the objective
function with respect to customer benefits, network losses, and
load levelling.
The critical analysis of MG EMSs based on linear and non-
linear programming approaches is summarized in Table 3. Cen-
tralized supervisor control architecture is used in all these MG
EMSs. The main contributions of the listed papers have been
earlier discussed in this subsection, where most of the carried
out researches were focused on energy resources optimization
in MG. However, much emphasis is needed to consider the ef-
fects of depth of discharge (DOD) on battery performance, ef-
fects of GHG emissions of CGs on environment, customer pri-
vacy issues, integration of DR, and system reliability.
4.1.2. EMS based on Dynamic Programming and Rule-based
Methods
Heymann et al. [75] proposed a Bellmans dynamic pro-
gramming method for optimal energy management of a stan-
dalone MG. The proposed EMS model includes operational
cost of CGs and load shedding cost in objective function. To re-
duce computational time of dynamic programming model, the
Pontryagin maximum principle is used. It determines five op-
timal extreme points using continuous and concave properties
of Hamiltonian instead of using all discretized control range.
The performance and efficiency is compared with classical non-
linear programming and MILP models in terms of operational
cost and computation time. The results show that the proposed
approach is more effective than the aforementioned methods.
In [76], an efficient EMS model is presented to optimize the
operation of a grid-connected MG. The objective is to mini-
mize the cash flow, which includes energy trading cost with
main grid and battery ageing cost. Two cases of constant en-
ergy and dynamic energy prices are studied. The proposed
dynamic programming approach performs better than the rule-
based method.
Strelec and Berka [77] have presented an approximate dy-
namic programming approach to overcome the curse of dimen-
sionality in the proposed EMS model of a grid-connected MG.
Receptive field weight regression and lookup table are used
to compute the approximate value of cost-function. Various
scenarios of wind speed, load demand, and ambient tempera-
ture are generated to consider their uncertainty in the proposed
model, which uses economic dispatch and unit commitment
operations to optimize the energy scheduling of MG. The ef-
fectiveness of the proposed approach is compared with myopic
optimization and dynamic programming methods. It gives bet-
ter results in reducing operating cost, but with higher computa-
tional time, as compared to myopic optimization method. How-
ever, it achieves less computation time than dynamic program-
ming method, but at the cost of higher objective function value.
A centralized rule-based energy management model is de-
veloped in [36] for both islanded and grid-connected modes
of MG and simulated in PSCAD/EMTDC software. In the is-
landed mode, fuel cell operates only if SOC of battery is below
80%. While, in grid-connected mode, SOC of battery should
remain more than 60% to have reliable operation in islanded
mode. This rule-based control model ensures smooth transition
between these two modes with respect to voltage and frequency
stability of MG system. Kanchev et al. [78] proposed a concept
of prosumers in an MG system. MG consists of prosumers and
a microturbine. A prosumer is composed of PV system, battery,
and ultra-capacitor. Two EMS models are considered for MG
operation. Central EMS uses rule-based optimization method
to control the energy operation of whole MG with an interval
of half an hour, while a prosumer EMS manages the power bal-
ancing and primary frequency regulation in prosumer system.
Merabet et al. [79] presented an online rule-based EMS to opti-
mize the operation of MG in real-time by switching operations
among battery charging mode, battery discharging mode, off-
MPPT mode of PV system, and load shedding mode on the
basis of load generation imbalance and SOC of battery.
Choudar et al. [80] presented a battery SOC-based hierar-
chical structure for MG EMS and proposed ultra-capacitors for
power regulation and smooth operation of MG. The SOC of
battery and energy level of ultra-capacitors determine the selec-
tion of four different operating modes, namely: normal mode,
PV limitation mode, recovering mode, and disconnection mode,
for efficient energy management of MG. A battery SOC-based
EMS for a grid-connected residential MG is recommended in
[81]. It uses persistence forecasting method for load demand
prediction and Meteogalicias THREDDS server forecasted data
for prediction of solar irradiance and wind speed. The central
moving average strategy is used to reduce peaks and fluctua-
tions within the energy exchange of MG with the main grid
through the battery SOC management. Sechilariu et al. [82]
proposed rule-based real-time EMS for building integrated MG.
The rule-based power control algorithm is presented to switch
operation among load shedding, battery charging and discharg-
ing, PV limitation, and grid supply modes. The operations are
selected on the basis of load generation imbalance, time of use
tariff, and battery SOC. To stabilize the DC bus voltage, the PI
controller is used to calculate the load generation imbalance us-
ing PV power, load demand, and reference DC bus voltage as
inputs. This power imbalance is compensated by battery supply
and main grid.
The critical analysis of MG EMSs based on dynamic pro-
gramming and rule-based approaches is summarized in Table 4.
These MG EMSs are mainly focused on energy resources opti-
mization and energy trading with main grid. However, a lot of
work is still required on mitigation of environmental pollutants,
consideration of DR, and inclusion of customer privacy issues
for MG systems optimizations.
7
Table 3: Critical analysis of MG EMSs based on linear and nonlinear programming methods.
Ref. Proposed
approach
Contributions Limitations Supervisory
control
Uncertainty
handling
approach
[35] Linear and mixed
integer linear
programming
Power sharing, continuous run and on/off-based mixed
mode EMS of MG is proposed.
Higher DOD of battery is selected that results
in its fast degradation
Centralized Forecasted
[29] Mixed integer
linear
programming
Net present cost is determined and advantages of DR
program are discussed with respect to load factor im-
provement and reduction in peak power consumption.
No CG is considered for effective management
of handling intermittency of RERs.
Centralized Forecasted
[66] Mixed integer
linear
programming
Fuel cost minimization model using piecewise linear
function of diesel generators and battery sizing problem
are proposed.
Uncertainties in PV power and load demand
are not considered.
Centralized Forecasted
[67] Mixed integer
linear
programming
Energy trading profit-based EMS model is proposed and
the feasible integration of thermal storage is realized.
Battery integration is discouraged. However,
no alternative solution is proposed to deal with
the intermittency of RERs.
Centralized Radial basis NN
[68] Mixed integer
linear
programming
DR and range anxiety term of EVs are considered while
optimizing daily energy resource scheduling of MG.
Effects of plug and play feature of EVs on MG
stability are not discussed.
Centralized Forecasted
[69] Mixed integer
linear
programming
Utility grid peak shaving incentives are considered in
optimal EMS model to maximize daily revenue of MG.
Assuming load to be always more than gener-
ation results in considerable load shedding and
consumer outrage during transmission lines
congestion and grid disturbances.
Centralized Forecasted
[47] Mixed integer
linear
programming
DR and penalty costs on load shedding are also included
in optimal EMS model of MG.
Battery degradation cost due to higher DOD is
ignored.
Centralized Forecasted
[70] Mixed integer
linear
programming
Outages constraint and load shedding cost are included
in three-phase EMS power flow model that is linearized
using approximation methods.
DR and power losses are not taken into ac-
count.
Centralized Forecasted
[71] Mixed integer
linear
programming
A comparative analysis is performed among the EMS
models with perfect forecast and imperfect forecast,
SOC-based EMS and EMS model with accurate infor-
mation in respect of savings and computational time.
Methods for predicting accurate information
related to MG system are not described. DR
is also not considered.
Centralized Forecasted
[72] Mixed integer
linear
programming and
nonlinear
programming
An optimal MG energy management model of an iso-
lated three-phase MG system is developed to minimize
the fuel, start-up and shutdown costs of CGs and penalty
cost on reactive power requirements.
Complex formulation. DR and voltage regula-
tion are not included.
Centralized Forecasted
[73] Mixed integer
nonlinear
programming
The operating cost of CGs is minimized, while consider-
ing droop controlled active and reactive power dispatch
of AC side CGs as a constraint.
Emission cost of CGs and DR incentives are
not considered.
Centralized Forecasted
[74] Nonlinear
programming
Market policies have been defined for MG participation
in energy market on the basis of its profit and operational
cost.
No ESS is considered in EMS model of MG. Centralized Forecasted
[37] Nonlinear
programming
Efficient MG EMS is proposed under operation window
of transformer nominal operation and voltage security.
Three objective functions of customer benefits, network
losses and load levelling are studied.
Operational cost of DERs is not considered.
Deep discharge battery with 100% DOD has
reduced lifecycle time and increased invest-
ment and replacement costs.
Centralized Forecasted
4.2. EMS based on Meta-Heuristic Approaches
4.2.1. EMS based on Genetic and Swarm Optimization
Elsied et al. [83] developed a multi-objective EMS using
Zigbee-based reliable communication infrastructure. It opti-
mizes the MG operation based on the objectives of minimiza-
tion of operation cost, minimization of GHGs emission cost,
and maximization of energy trade profit. Economic load dis-
patch and battery degradation cost-based multi-objective EMS
of a remote MG is proposed in [84]. It performs both day-ahead
and real-time operations using genetic algorithm (GA) and rule-
based approach, respectively. The real time operation consid-
ers diesel generator supply, battery supply, and load shedding
options in a sequential order to maintain load generation bal-
ance. Askarzadeh [34] proposed a memory-based GA for op-
timal power management of a grid-connected MG to minimize
the operating cost of DERs. The performance of the proposed
approach is better as compared to GA, particle swarm optimiza-
tion (PSO) with inertia factor and PSO with constriction factor.
Chen et al. [65] proposed ESS economical model and ma-
trix real coded GA-based smart EMS model for a grid-con-
nected MG. An NN model is used to predict solar power. The
ESS economical model maximizes the net present cost of ESS
over its lifespan based on its capital, operation and maintenance
costs, and energy arbitrage revenue. The smart EMS model
incorporates generator bids, storage bids, and energy trading
profit to minimize the operational cost of MG while satisfy-
ing energy balance constraint and physical constraints of DERs.
Golshannavaz et al. [85] presented a GA-based EMS for opti-
mal generation and reserve scheduling of a grid-connected MG.
It models uncertainties of wind power and load demand by sce-
nario generation and reduction method. The objective func-
tion includes the operating cost, energy purchase cost, reactive
power support cost, active power reserve cost, demand respon-
sive cost, switching cost of automatic controlled switches, and
load shedding cost. Automatic controlled switches are used to
enhance the techno-economic performance of MG. Radosavl-
8
Table 4: Critical analysis of MG EMSs based on dynamic programming and rule-based methods.
Ref. Proposed
approach
Contributions Limitations Supervisory
control
Uncertainty
handling
approach
[75] Dynamic
programming
Operational cost of CGs and penalty cost on load shed-
ding are considered in objective function. To reduce
computational time, the Pontryagin Maximum Principle
is used.
Higher DOD is selected that leads to fast
degradation of battery. DR is not considered.
Centralized Forecasted
[76] Dynamic
programming
Cash flow, which includes energy trading cost with main
grid and battery ageing cost, is minimized.
Computational time complexity of proposed
approach is not discussed.
Centralized Forecasted
[77] Approximate
dynamic
programming
Approximate value of cost-function in economic dis-
patch and unit commitment operations is considered to
optimize the daily energy scheduling of MG.
DOD of battery for MG optimized operation
with battery ageing status are not discussed.
Centralized Markov chain
[36] Rule-based
approach
In an islanded mode, fuel cell operates only if battery
SOC is below 80%. In grid-connected mode, battery
SOC remains more than 60% to ensure reliable opera-
tion in islanded mode.
DR is not included. The model lacks the con-
sideration of future availability of RERs.
Centralized Forecasted
[78] Rule-based
approach
Central EMS controls the energy operation of whole
MG, while prosumer EMS manages the power imbal-
ancing in prosumer system.
GHG emissions cost and DR are not included. Centralized Forecasted
[79] Rule-based
approach [79]
EMS switches operation among different modes on the
basis of load generation imbalance and SOC of battery.
No DR and knowledge of future availability
profiles of RERs and load demand.
Centralized Forecasted
[80] Battery SOC
rule-based
approach
A battery SOC-based hierarchical structure for MG
EMS is proposed together with ultra-capacitors for
power regulation and smooth operation of MG.
No prior information of generation of RERs.
DR and voltage regulation of MG system are
not taken into consideration.
Centralized Forecasted
[81] Battery SOC
rule-based
approach
Fluctuations in power transaction with the main grid are
reduced.
Higher DOD is selected that leads to fast
degradation of battery.
Centralized Persistence
forecasting
method
[82] Rule-based
approach
Power balancing of building integrated MG is achieved.
DC bus voltage stabilization is also ensured.
Operation cost, energy transaction cost and DR
are not considered.
Centralized Forecasted
jevi et al. [86] proposed an optimal EMS for a grid-connected
MG that considers uncertainties of RERs, load demand, and
electricity price using point estimate method. The efficiency of
PSO in finding best solution is shown to be better in compari-
son with GA, combinatorial PSO, fuzzy self-adaptive PSO, and
adaptive modified PSO.
Li et al. [87] proposed a regrouping PSO-based optimal en-
ergy management strategy for industrial MG that can be oper-
ated islanded and grid-connected modes. In the islanded mode,
the objective function aims to minimize the O&M costs of MG.
However, for grid-connected mode, it also maximizes the en-
ergy trading profit with the main grid. The performance of the
proposed approach is better as compared to GA in respect of
global optimum solution and computation time. A novel guar-
anteed convergence PSO with Gaussian mutation algorithm is
proposed in [31] for optimal operation of a standalone MG.
The objective function is composed of capital and O&M costs
of MG. The proposed algorithm has been implemented on 69-
bus and 94-bus isolated MG system that performs better than
GA and PSO algorithms. Alavi et al. [88] developed an op-
timal EMS model to optimize the operation of MG. Beta and
Weibull probability density functions are used in point esti-
mated method to model the uncertainties of solar power and
wind power, respectively. However, uncertainties in load de-
mand are modeled by robust optimization approach. The ob-
jective of the developed EMS model is to minimize the O&M
costs, emission cost, and reliability cost of MG.
Baziar and Kavousi-Fard [40] proposed a novel self-adap-
tive modified θ-PSO method to minimize the operating cost of
MG. Point estimated method is used to model the uncertain-
ties of RERs, load demand, and electricity price. Two scenarios
are studied in context of availability of distributed generators
and battery initial SOC. The proposed algorithm performance
is better in comparison to GA, combinatorial PSO, fuzzy self-
adaptive PSO, adaptive modified PSO and θ-PSO. Multi-objec-
tive PSO and advanced metering infrastructure-based EMS for
an isolated AC/DC MG is presented in [89]. Operation value
factor for each objective is introduced in the proposed optimal
operation model of MG that has been defined by grid opera-
tors on the basis of day-ahead analysis. The objective function
is composed of battery operation value, volt-var optimization,
generation reduction, load shedding, and power transfer value
between AC and DC MGs. The performance of the proposed
solution is tested on 33-node MG system that shows the im-
proved performance in achieving optimal solution.
Moghaddam et al. [48] proposed an adaptive modified PSO
approach based on hybridization of chaotic PSO and fuzzy self-
adaptive PSO to optimize the multi-objective EMS model of a
grid-connected MG. The objectives is to minimize operational
and emission costs of MG. The developed algorithm performs
better than GA, PSO, chaotic PSO, and fuzzy self-adaptive PSO.
Mohan et al. [90] proposed an energy and reserve management
system for optimal operation of an MG. The objectives are to
minimize the fuel cost, emission cost, load shedding cost, volt-
age deviation, active and reactive power mismatches, and en-
ergy trading cost with the main grid. The proposed approach is
solved by affine arithmetic and stochastic weight tradeoffPSO-
based perturbed optimal power flow method, and it performs
better than interval arithmetic method.
Table 5 presents the critical analysis on MG EMSs that are
solved by genetic and swarm optimization approaches. Many
objectives deserve to be explored in detail such as the compu-
9
tational complexity of the proposed approaches, battery degra-
dation cost, effects of GHG emissions of CGs on environment,
and reliable and sustainable operation of MGs during islanded
mode. Scenario generation and reduction method, point esti-
mate method, and affine arithmetic methods are used to con-
sider uncertainties within an MG system.
4.2.2. EMS based on other Meta-Heuristic Approaches
In [91], a differential evolution approach is presented for
optimal energy management of a grid-connected MG. The ob-
jectives are minimization of operational and emission costs of
MG that have been optimized separately. Operational cost of
MG includes bidding cost of DERs, DR incentives, and energy
trading cost with main grid. The effects of DR in reducing peak
shave demand and GHG emissions are also studied. The ef-
ficiency and efficacy of the proposed algorithm is better than
PSO in terms of best solution and convergence speed. Yu et
al. [92] proposed an optimal EMS model of a grid-connected
MG, which determines the energy scheduling of MG with the
objective of its optimal economic operation. The economical
objectives are profit on selling energy to load end and main
grid, energy purchasing cost with main grid, and battery ageing
cost. The proposed approach is more efficient than rule-based
method in achieving best economic operation of MG.
In [93], ant colony optimization-based two-layer EMS mod-
el for an islanded MG is proposed to minimize its operational
cost. It includes bidding cost of RERs, CGs and battery, penalty
cost on load shedding, and DR incentives in both day-ahead
scheduling and 5 minutes interval real-time scheduling layers.
Three scenarios of normal operation, sudden high requirement
of load demand, and plug and play ability are studied with
experimental validation. The proposed approach reduces op-
erational cost of MG for almost 23% and 5% more than the
modified conventional EMS and PSO-based EMS, respectively.
Marzband et al. [94] introduced the gravitational search al-
gorithm-based EMS model of an isolated MG that minimizes
the operating cost of MG together with penalty cost on unde-
livered power. Three scenarios of normal operation, sudden
high requirement of load demand, and plug and play feature
are taken into consideration to evaluate the effectiveness of the
proposed approach with experimental validation. The proposed
algorithm performance is better as compared to PSO.
Niknam et al. [95] proposed a probabilistic EMS to min-
imize the operational cost of a grid-connected MG. The 2m
point estimate method is used to model MG uncertainties. The
developed optimal approach is solved by self-adaptive gravi-
tational search algorithm that performs better than GA, PSO,
fuzzy self-adaptive PSO, and gravitational search algorithm. A
modified bacterial foraging method-based multi-objective EMS
model of a grid-connected MG is presented in [43]. It mini-
mizes the objectives of operational and GHG emission costs.
The sensitivity analysis on tradeoffbetween these two objec-
tives is also studied using interactive fuzzy satisfactory method.
The performance of the modified bacterial foraging method is
shown to be better as compared to GA and PSO.
Motevasel and Seifi [96] proposed an expert EMS for opti-
mal operation of a grid-connected MG that minimizes conflict-
ing objectives of operation and emission costs. Artificial NN is
used to forecast the wind speed. The proposed algorithm per-
forms better than GA and PSO in respect of determining best
solution. Marzband et al. [97] proposed multi-period artifi-
cial bee colony-based experimental two-layer EMS model for
residential MG. It minimizes operational cost that includes lin-
earized operating cost of RERs, CGs and battery, penalty cost
on load shedding and incentives for controllable load in first
layer of day-ahead scheduling. Similarly, second layer, 5 min-
utes interval real-time scheduling, considers operating cost of
RERs, CGs and battery, penalty cost on load shedding, power
imbalance, energy trading cost with main grid, and incentives
for DR. The proposed solution approach is concluded to be
more productive than mixed integer nonlinear programming al-
gorithm. An EMS solution approach for an isolated MG is pro-
posed in [98] to minimize its operational cost that includes op-
erating cost of RERs, microturbine and battery, penalty cost on
load shedding, and DR incentives. A hybrid NN and markov
chain-based prediction method is used to forecast power gen-
eration of RERs and load demand. The proposed algorithm is
more effective than PSO in respect of computation time and op-
timum solution.
Mohammadi et al. [99] presented a scenario-based EMS
model for a grid-connected MG. It utilizes adaptive modified
firefly algorithm to minimize operational cost considering un-
certainties of RERs, load demand, and electricity price. Sce-
nario generation and reduction method is used to determine
these uncertainties by using their probability density functions.
The performance of the proposed algorithm is better than PSO,
fuzzy self-adaptive PSO, GA, firefly algorithm with respect to
computational time, and finding best optimum solution. A multi-
objective EMS of a grid-connected prosumer building MG, whi-
ch is composed of PV system and battery, is presented in [100].
It aims to maximize MG profit by trading energy with main
grid and neighboring building MGs. It also ensures that load
demand is always met and PV produced power is not wasted.
Arefifar et al. [45] proposed a new probabilistic index, en-
ergy management success index, as an objective function to re-
duce the total operational cost of multi-MG system using tabu
search method. The optimization model includes multi-state
variable concept to consider the uncertainties of RERs and EVs.
Selection of the number of states for each uncertain variable
depends on the requirements of processing time and improve-
ment in optimum solution of developed optimization problem.
The forward-backward-based probabilistic power flow method
is used to determine the energy losses of the system. In [101],
a modified crow search-based EMS model was presented to op-
timize the daily energy scheduling of a grid-connected MG.
The objective is to minimize the operational cost of MG that
includes bidding costs of DERs and energy trading cost with
main grid. Three scenarios are studied with respect to opera-
tional status of distributed generators and value of battery ini-
tial SOC. The proposed approach is more efficient and effective
than GA, PSO, fuzzy self-adaptive PSO, and crow search al-
gorithm in respect of best solution and reduced computational
time.
In [102], the overall operational cost of MG is minimized. It
10
Table 5: Critical analysis of MG EMSs based on genetic and swarm optimization.
Ref. Proposed
approach
Contributions Limitations Supervisory
control
Uncertainty
handling
approach
[83] Genetic
Algorithm
A multi-objective EMS model considers operation cost,
emission cost and energy trade profit as objectives for
optimal operation of MG.
DR and DOD of battery are not considered.
The computational time complexity of the pro-
posed approach is not determined.
Centralized Forecasted
[84] Genetic algorithm Battery degradation cost is considered to increase its
lifecycle together with the operational cost model of
MG.
Emission cost of diesel generator is not taken
into consideration.
Centralized Forecasted
[34] Memory-based
genetic algorithm
Quadratic cost-functions of DERs are considered in pro-
posed EMS model to determine their optimal schedul-
ing.
No ESS is taken into consideration and com-
putational time complexity is not calculated.
Centralized Forecasted
[65] Matrix real coded
genetic algorithm
DERs marginal price bids are proposed in smart EMS
model to minimize the operational cost of MG.
Emission cost of microturbine and DR are not
taken into consideration.
Centralized Artificial NN
[85] Genetic algorithm Optimal scheduling of active and reactive power, and re-
serve power are determined. The cost related to load
shedding and automatic controlled switches are also
considered.
Power losses and ESSs are not taken into con-
sideration.
Centralized Scenario
generation method
[86] Particle swarm
optimization
Operation cost of MG is minimized. The point estimated
method is used to model uncertainties of RERs, load de-
mand and electricity price.
Emission cost of diesel generator and DR are
not considered.
Centralized Point estimate
method
[87] Regrouping PSO Global optimum solution is computed considering oper-
ational cost and energy trading cost of an industrial MG.
DR and emission cost of diesel generators are
not considered.
Centralized Forecasted
[31] Guaranteed
convergence PSO
A novel PSO variant is developed and implemented on
69-bus and 49-bus MG system to validate its perfor-
mance in respect of optimal operation of MG.
Authors have not considered emission cost of
diesel generator and power losses of MG sys-
tem.
Centralized Forecasted
[88] Particle swarm
optimization
Operation, emission and reliability costs of MG are opti-
mized. Solar and wind powers uncertainties are modeled
by point estimate method.
Complex formulation of uncertainty modeling.
Higher DOD leads to fast degradation of bat-
tery lifetime. Power losses are also not consid-
ered.
Centralized Point estimate
method
[40] self-adaptive
modified θ-PSO
A novel optimization algorithm is proposed to minimize
the operating cost of MG. The point estimated method
is used to model MG uncertainties.
Lifecycle cost of battery and DR are not con-
sidered.
Centralized Point estimate
method
[89] Multi-objective
PSO
Grid operator-based operational value factor is defined
for each objective function to maximize revenue of MG.
Lifecycle cost of battery and emission cost of
diesel generators are not scrutinized.
Centralized Forecasted
[48] Adaptive modified
PSO
A multi-objective operation model of MG is proposed to
minimize its operational and emission costs.
Computational time complexity of the pro-
posed approach is not discussed. DR is not
taken into consideration.
Centralized Forecasted
[90] stochastic weight
tradeoffPSO
The fuel cost, emission cost, load shedding cost, voltage
deviation, active and reactive power mismatches, and
energy trading cost with the main grid are minimized.
Complex formulation. Power losses cost and
reliability of MG system are not discussed. DR
is also not taken into consideration.
Centralized Affine arithmetic
includes fuel cost of CGs, operating cost of DERs, energy trad-
ing cost with main grid. The least square support vector ma-
chine method is used to predict the power output of RERs. An
EMS solution approach for a standalone MG based on imperial-
ist competition algorithm is proposed in [103] to minimize the
operational cost of MG together with penalty cost on undeliv-
ered power. A hybrid NN and Markov chain-based prediction
method forecast power generation of RERs and load demand.
Three scenarios of normal operation, sudden high requirement
of load demand, and plug and play ability are taken into con-
sideration. The effectiveness of the proposed approach is also
validated experimentally. It is proven that the proposed algo-
rithm performs better than PSO.
Table 6 summarizes the critical analysis on MG EMSs that
are solved by meta-heuristic approaches, other than genetic and
swarm optimization. In these approaches, a lot of work is still
required on considering joint reduction of operational cost and
environmental pollutants, reduction of MG systems losses, inte-
gration of DR, and reliability improvement. Centralized super-
visor control architecture is used in all these MG EMSs. Few
authors also used efficient uncertainty quantification methods
such as scenario generation and reduction method, and point
estimate method to take into account the effects of MG uncer-
tainties in achieving a reliable and sustainable operation.
4.3. EMS based on Artificial Intelligent Methods
4.3.1. EMS based on Fuzzy Logic and Neural Network
Arcos-Aviles et al. [104] presented a fuzzy logic-based
EMS to achieve smooth power profile of a grid-connected res-
idential MG. The proposed model minimizes the fluctuations
and power peaks in the energy exchange with the main grid. It
also maintains the battery SOC level close to 75% of its rated
capacity to improve its lifetime. The proposed approach per-
forms better than SOC-based EMS. Chen et al. [105] intro-
duced an efficient EMS model for a standalone DC MG using
fuzzy logic control and Zigbee-based communication network.
The proposed model ensures efficient utilization of RERs and
improvement in lifetime of Li-ion battery. The efficacy and ef-
ficiency of the proposed approach is also experimentally vali-
dated. Kyriakarakos et al. [106] proposed a fuzzy logic-based
EMS for an isolated MG that minimizes net present cost to-
gether with penalty cost on battery SOC, hydrogen, and water
storages. The load demands are divided into three categories of
electric load, water load, and transport load. The hydrogen is
11
Table 6: Critical analysis of MG EMSs based on other meta-heuristic methods.
Ref. Proposed
approach
Contributions Limitations Supervisory
control
Uncertainty
handling
approach
[91] Differential
Evolution
Operational and emission costs of MG are minimized
separately, while considering DR.
Battery degradation cost, in terms of DOD, is
not considered.
Centralized Forecasted
[92] Modified
differential
evolution
Economic operation of MG, which also considers bat-
tery ageing cost, is optimized.
DR is not considered. Centralized Forecasted
[93] Ant colony
optimization
Operational cost of MG, which includes bidding cost of
DERs, penalty cost on load shedding and DR incentives,
is minimized.
Battery degradation cost and emission cost of
CGs are not taken into consideration.
Centralized Forecasted
[94] Gravitational
search algorithm
Operating cost of an isolated MG is minimized consid-
ering three scenarios of normal operation, peak load de-
mand, and plug and play ability.
High DOD leads to fast degradation of battery
lifetime. Emission cost of CG is not consid-
ered.
Centralized Forecasted
[95] Self-adaptive
gravitational
search algorithm
An EMS strategy is developed to minimize operational
cost of an MG together with 2m point estimate method
to model its uncertainties.
Emission cost of microturbine and power
losses of MG system are ignored.
Centralized Point estimate
method
[43] Modified bacterial
foraging
A multi-objective intelligent EMS is proposed to mini-
mize the operation and emission costs of MG.
High computational time complexity. DR is
not considered.
Centralized Forecasted
[96] Modified bacterial
foraging
An expert EMS is proposed to minimize the operation
and emission costs of MG. Artificial NN is used to fore-
cast wind speed.
High computational time complexity. DR is
not taken into consideration.
Centralized Artificial NN
[97] Artificial bee
colony
Operational cost of home MG is minimized based on
two layer control model with experimental validation.
Complex formulation. Emission cost of micro-
turbine is ignored.
Centralized Forecasted
[98] Modified artificial
bee colony
DR and load shedding are also inspected together with
operating costs of DERs to minimize operational cost of
a standalone MG.
Lifecycle cost of battery and emission cost of
microturbine are not taken into consideration.
Centralized Forecasted
[99] Adaptive modified
firefly
A scenario-based EMS model for a grid-connected MG
is introduced to minimize its operational cost consider-
ing MG uncertainties.
Emission cost of microturbine and reliability
cost in terms of undelivered power are not con-
sidered.
Centralized Scenario
generation and
reduction method
[100] Modified
simulated
annealing
Energy trading profit with main grid and neighboring
MGs is maximized and it is also ensured that load de-
mand must be always satisfied and PV power must not
be wasted.
Selection of DOD of battery is not discussed.
DR is not taken into consideration.
Centralized Forecasted
[45] Tabu search A new probabilistic index, energy management success
index, is defined as an objective function to reduce the
total operational cost and power losses of the MG sys-
tem.
Computational time complexity of proposed
approach is not discussed.
Centralized Scenario
generation method
[101] Modified crow
search algorithm
The operational cost of MG, which includes bidding cost
of DERs and energy trading cost with main grid, is min-
imized.
DR and emission costs of CGs are not consid-
ered.
Centralized Forecasted
[102] Modified artificial
fish school
algorithm
The overall cost of MG, which includes operating cost
of DERs and energy trading cost with main grid, is min-
imized to optimize its daily operation.
Emission cost of CG is not considered. Centralized Least square
support vector
machine
[103] Imperialist
competition
algorithm
Operational cost of a standalone MG is minimized. A
hybrid artificial NN and Markov chain method are used
to forecast RERs and load demand
High DOD leads to fast degradation of battery
lifetime. Emission cost of CG is not consid-
ered.
Centralized Forecasted
considered as a fuel for transport load. The decision inputs for
the fuzzy logic system are battery SOC, water, and system fre-
quency. The MG system simulation is performed for one year
study on software platforms of Transys, Genopt, Matlab, and
Trnopt.
Enrico et al. [107] introduced a fuzzy logic-based EMS of
a grid-connected MG. The hierarchical genetic algorithm tunes
Mamdani fuzzy inference systems to minimize the fuzzy rules
of an EMS model. The improved MG model is realized by tak-
ing into account the realistic efficiency parameters of battery
that replaces an ideal battery model. The developed approach
optimizes the fuzzy rule base of an EMS for efficient energy
flow and maximization of energy trading profit. The perfor-
mance of the proposed approach appears to be better as com-
pared to a classical fuzzy-GA method. Fuzzy-based MG eco-
nomic dispatch and unit commitment is performed in [39] that
considers two GAs to optimize its energy scheduling operation.
First GA determines MG energy scheduling and fuzzy rules,
while the second GA tunes fuzzy membership functions. Fuzzy
expert system is also used to manage the power allocation of
battery.
Chaouachi et al. [49] proposed an efficient multi-objective
MG EMS to minimize the operational and emission costs of
MG, which includes charging and discharging rates of battery
determined by fuzzy expert system. The artificial NN ensemble
method forecasts the power generation of RERs and load de-
mand. The efficiency of the proposed approach is better com-
pared to conventional multi-objective model that does not con-
sider fuzzy expert system for battery scheduling. Wang et al.
[108] presented a Lagrange programming NN approach for ef-
ficient MG EMS to minimize overall cost of MG that includes
fuel cost, operation and maintenance costs, and emission cost of
generation units. A radial basis NN forecasts power generation
of RERs and load demand. Load is divided into four categories
12
of critical load, controllable load, price sensitive load, and ther-
mal load to consider DR. The proposed approach achieves the
best solution more efficiently than PSO. Urias et al. [109] in-
troduced a recurrent NN approach for efficient EMS of a grid-
connected MG. It aims to minimize power import from main
grid and maximize utilization of power output of RERs. DOD
of battery is set to 60% to improve its lifetime. Load demand is
divided into two categories namely: critical and ordinary loads.
A hybrid wavelet functions and extended Kalman filter-based
NN approach is adopted to forecast the load demand and power
generation of RERs.
Venayagamoorthy et al. [110] proposed an intelligent adap-
tive dynamic EMS for a grid-connected MG. It maximizes the
utilization of RERs and minimizes carbon emissions to achieve
a reliable and self-sustainable system. It also improves bat-
tery lifetime. The proposed EMS is modeled using evolution-
ary adaptive dynamic programming and reinforcement learning
concepts and solved by use of two NNs. An active NN is used
to solve the proposed EMS strategy, while a critical NN checks
its performance with respect to optimality. The new defined
performance index evaluates the performance of dynamic EMS
in terms of battery lifetime, utilization of renewable energy, and
minimum curtailment of controllable load. The performance of
the proposed approach is better as compared to decision tree ap-
proach-based dynamic EMS. Kuznetsova et al. [111] proposed
a reinforcement learning-based MG EMS to achieve the maxi-
mum utilization of battery and wind energy resources, and the
minimum dependence on energy purchase from main grid. In
reinforcement learning, a Q-learning method is used to achieve
the optimal operation of MG. To account wind speed uncer-
tainties, Markov chain model is used to generate scenarios for
forecasted wind speed. Two hours-ahead scenarios are consid-
ered only in optimization process to realize low computational
cost of forecasting method.
Table 7 summarizes the critical overview on MG EMSs that
are solved by fuzzy logic and NN methods. Centralized super-
visory control architecture is used in all these MG EMSs. The
uncertainties within an MG system are taken into account by
using forecasted values. Most of the contributions are DERs
operational cost in minimization, mitigation of environmental
pollutants, and energy trading with main grid. However, com-
putational complexity of the proposed approaches, customer
privacy issues, DR, reliability of MG system, and losses cost
of MG are not yet assessed in detail.
4.3.2. EMS based on Multi-Agent System
A MAS-based decentralized approach for optimal operation
of a grid-connected MG is presented in [112]. All the con-
sumers, storage units, generation units, and grid are considered
as agents. In the decision making, the consumer consumption
preference has been considered as an important factor. The
multi-agent decentralized algorithm reduces the power imbal-
ance cost while considering consumer consumption preference
as an important factor in the decision making process. The au-
thors concluded that the decision making time of the proposed
approach is better than the centralized approach.
Bogaraj and Kanakaraj [41] presented a MAS-based ap-
proach to design an intelligent EMS and load shedding scheme
for an islanded MG. It maintains energy balance through ef-
fective coordination among RERs, batteries, and loads. The
agents are photovoltaic system, wind turbine, fuel cell, battery,
and load. Fuel cell and battery bank are used as a backup. Load
is divided into three groups based on priority. The power gen-
eration of RERs, loads, and atmospheric temperature, are fore-
casted using auto regressive moving average (ARMA) method.
Two case studies of high wind and high irradiance, and low
wind and low irradiance, are used to check the performance of
the proposed approach. It also minimizes time delays in mak-
ing decisions. STATCOM is used for reactive power compen-
sation to reduce harmonics and improve voltage profile of MG
system. Anvari-Moghaddam et al. [113] proposed an ontology
driven MAS-based EMS for efficient operation of a grid-con-
nected residential MG. The agents are categorized into central
coordinator agent, building management agent, RER agent, bat-
tery agent, and service agent. The objectives of the proposed
approach are to minimize operational cost and to satisfy con-
sumers electrical and thermal comfort levels. Intra agent plat-
form communication uses internal message transport protocol
(IMTP), while inter agent platform communication uses hyper-
text transfer protocol (HTTP).
Nunna and Doolla [114] proposed an optimal MG EMS,
considering DR and distributed storage, that aims to reduce
peak demand and minimize electricity cost. Consumers are en-
couraged to participate in DR program by using an index-based
incentive mechanism for them. The agents are generation agent,
storage agent, load agent, DR agent, MG intelligent agent, and
global intelligent agent. The MAS architecture is implemented
on Java application development framework. Two case studies
of two and four interconnected MGs are performed after each
15 minutes interval for day-ahead operation. Dou and Liu [115]
proposed a multi-objective hierarchical MAS-based decentral-
ized EMS for a grid-connected MG system to minimize its op-
erational cost, emission cost, and line losses. The hierarchical
MAS is divided into three levels. The upper level agent deals
with the energy optimization of MG. The middle level agents
are concerned with the coordination among control agents to
switch operation modes using Petri-net model for voltage reg-
ulation. The lower level agents define f/Vand PQ-based con-
trol strategies for unit agents to manage real-time operation of
DERs.
A fuzzy cognitive maps-based MAS approach is proposed
in [116] for decentralized EMS of an islanded MG. It mini-
mizes system net present cost together with penalty cost on bat-
tery SOC, hydrogen storage, and water storage. Five intelli-
gent agents are defined, which are RER agent, battery agent,
desalination agent, fuel cell agent, and electrolyzer agent, re-
spectively. The proposed MAS-based approach is simulated by
interconnection of Transys, Matlab, Genopt and Trnopt soft-
ware packages and is compared with fuzzy logic-based central-
ized EMS method in terms of economic benefits.
Table 8 summarizes the critical analysis on the contribu-
tions and main limitations of MG EMSs using multi-agents, to-
gether with the supervisory architectures. In this context, com-
13
Table 7: Critical analysis of MG EMSs based on fuzzy logic and neural network methods.
Ref. Proposed
approach
Contributions Limitations Supervisory
control
Uncertainty
handling
approach
[104] Fuzzy logic Fluctuations and power peaks are minimized while ex-
changing energy with the main grid.
Voltage and frequency regulation of MG sys-
tem is not considered.
Centralized Forecasted
[105] Fuzzy logic Lifetime of Li-ion battery is improved. Zigbee is used
for communication among MG components.
It is difficult to meet load demand always due
to no DR involvement and battery SOC level to
be set 50% as a threshold.
Centralized Forecasted
[106] Fuzzy logic Penalty cost on battery SOC, water storage, and hydro-
gen storage are included in EMS model to minimize
overall operational cost of MG.
DR is not considered. The computation time
complexity of the proposed approach is not
discussed.
Centralized Forecasted
[107] Fuzzy logic A realistic battery parameters are included in MG
model. Energy exchange profit is maximized.
Complex formulation that results in higher
computational time.
Centralized Forecasted
[39] Fuzzy logic Fuzzy-based EMS approach is proposed for MG
scheduling and battery power management.
Higher computational time complexity. Power
losses are not considered.
Centralized Forecasted
[49] Fuzzy logic A multi-objective MG EMS minimizes the operation
and emission costs of MG that utilizes charging and dis-
charging rates of battery determined by fuzzy expert sys-
tem.
Computational time complexity of proposed
approach is not discussed. DR is not taken into
consideration.
Centralized Artificial NN
ensemble
[108] Neural network Overall cost of MG is minimized, radial basis NN is used
for prediction, and results are compared with PSO.
Computational time complexity of proposed
approach is not discussed. Operational cost of
battery is not considered
Centralized Radial basis NN
[109] Recurrent neural
network
Power import from main grid is discouraged to utilize
maximum renewable power availablewithin an MG sys-
tem.
Computational time complexity of proposed
approach is not discussed. DR is not consid-
ered.
Centralized Extended Kalman
filter-based NN
[110] Neural Network Maximum utilization of RERs and DR, improvement in
battery lifetime, and reduction in carbon emissions are
achieved.
Complex formulation. Computational time
complexity of the proposed approach is not
discussed.
Centralized Forecasted
[111] Reinforcement
learning
based-NN
MG operational performance is optimized by maximiz-
ing use of battery and RER, and minimizing energy pur-
chase from main grid.
Complex formulation. DR is not considered. Centralized Markov chain
putational time complexity of the proposed approach, reduced
GHG emissions, secure and reliable communication system for
decentralized operation of MG, effects of selection of DOD on
battery and MG performance, minimization of outages and in-
terruptions, and use of efficient uncertainty quantification meth-
ods are not addressed. These potential areas should be assessed
in detail to achieve energy efficient and environmental friendly
operation of MG system.
4.3.3. EMS based on other Artificial Intelligent Methods
Liu et al. [117] proposed stackelberg game theory approach
for optimal energy management and energy sharing of PV pro-
sumers-based grid-connected MG. It works on the principle of
leader-follower model, where MG acts as a leader and all PV
prosumers are followers. In the proposed model, the uncer-
tainties of solar power and load demand are tackled by a billing
mechanism approach. In this approach, the hourly deviation be-
tween scheduled and real power is compensated by regulating
revenues or costs of PV prosumers with respect to their con-
tributions in MG profit. The leader objective is global profit
maximization, while follower objectives are maximum utiliza-
tion of energy resources on prosumers level. However, a pro-
sumer with higher preference parameter in consuming its en-
ergy has higher utilization level which, in turn, decreases profit.
Therefore a tradeoffexists between utilization level and profit
in reference to preference parameter.
Ma et al. [118] presented a leader follower-based game the-
oretic EMS for a grid-connected MG with DR mechanism in
which MG and prosumers are considered leader and follow-
ers, respectively. The proposed approach aims to maximize the
profits of MG and prosumers individually, while maintaining
Stackelberg equilibrium to ensure fair share of profit distributed
among them. The MG operator uses differential evolution and
prosumers use nonlinear programing method to reach the Stack-
elberg equilibrium. The sensitivity analysis with respect to in-
creasing number of PV prosumers shows upward trend in MG
profit and good convergence rate of the proposed approach. It
performs better than a centralized optimization model. In [119],
a bi-level leader follower programming method is developed
for optimal energy management of multi-MGs. The upper level
deals with minimization of production cost of the whole sys-
tem. Lower level is related to maximization of net profit of each
MG. To reduce the effects of forecast errors of solar and wind
powers, the proposed model is solved for a short duration, of
the order of a few minutes to half an hour. The bi-level decen-
tralized problem is converted into single level problem by using
Karush-Kuhn-Tucker method for lower level problem, which is
solved by LINDOGlobal package in GAMS software.
A modified game theory-based multi-objective EMS of a
grid-connected MG is proposed in [120] to minimize Pareto
optimal objectives that include operation and emission costs of
MG. In the proposed approach, the objective function is defined
as the difference of Pareto objectives and supercriterion. Super-
criterion is the normalized distance of the objective function
from its worst value with 0 being best and 1 being worst. The
developed multi-objective optimization model is compared in
performance with multi-objective GA, multi-objective Sequen-
tial Quadratic Programming, and multi-objective Mesh Adap-
tive Direct Search. Lan et al. [121] introduced a rolling horizon
Markov decision process-based EMS of a grid-connected MG.
14
Table 8: Critical analysis of MG EMSs based on multi-agent system.
Ref. Proposed
approach
Contributions Limitations Supervisory
control
Uncertainty
handling
approach
[112] Multi-agent
system
Power imbalance cost is minimized, while considering
consumer consumption preferences in the decision mak-
ing process.
Effects of charging and discharging cycles on
batter lifetime are ignored.
Decentralized Forecasted
[41] Multi-agent
system
Load generation balance is achieved through battery
scheduling and load shedding. STATCOM is used to
improve power quality of MG.
Operational cost of RERs and battery are ig-
nored. Penalty cost on load shedding is also
not considered.
Decentralized Auto regressive
moving average
[113] Multi-agent
system
Multi-objective EMS is proposed to minimize opera-
tional cost of MG and to maximize consumers electrical
and thermal comfort levels.
Emission cost of microturbine is ignored. Op-
erational cost of battery is also not considered.
Decentralized Forecasted
[114] Multi-agent
system
Peak demand and electricity cost are minimized using
DR and distributed storage.
Higher DOD is selected that leads to fast
degradation of battery.
Decentralized Forecasted
[115] Multi-agent
system
Multi-objective formulation of EMS is proposed to min-
imize operational cost, emission cost, and line losses of
MG.
Complex formulation. Computation time com-
plexity of proposed approach is not discussed.
Decentralized Forecasted
[116] Multi-agent
system
The net present cost of MG system and penalty cost on
battery SOC, hydrogen storage, and water storage are
minimized.
DR is not considered. The computation time
complexity of the proposed approach is not
discussed.
Decentralized Forecasted
It minimizes electricity and natural gas costs considering wind
power uncertainties modeled by Markov decision process. The
greedy algorithm finds the feasible base policy and local optima
for rollout algorithm that solves the developed EMS model to
tackle large states of Markov decision process. The proposed
approach is more efficient than scenario tree-based method in
terms of both performance and computation time.
Jia et al. [122] presented an adaptive intelligence technique
for the EMS of a grid-connected MG using hybrid ESS. It aims
to maximize utilization of RERs and minimize load fluctuations
considering dispatch power errors due to uncertainties in power
generation of RERs and load demand. The smooth load profile
is managed by joint operation of battery and ultra-capacitors,
where sudden requirements of load demand are handled by ul-
tra-capacitors. The hybrid ESS efficiency, ratio of energy dis-
charge to the available energy, is higher in the proposed method
as compared to PSO.
Table 9 summarizes the critical analysis on contributions
and main limitations on MG EMSs based on other artificial
intelligent techniques such as game theory and Markov deci-
sion process together with the supervisory architectures of these
EMSs and uncertainty quantification methods. In these solution
approaches, much emphasis is needed to consider communi-
cation system cost for decentralized operation of MG, mitiga-
tion of environmental pollutants, losses cost, EVs integration
for DR, and system voltage and frequency regulation to ensure
reliable and sustainable operation of MG. Computational time
complexity of the proposed approaches have not been dealt with
too.
4.4. EMS based on Stochastic and Robust Programming Ap-
proaches
Ghasemi [123] proposed a grid-connected community MG
architecture for agriculture purposes. The author introduced
a stochastic coordination framework to minimize the cost of
pumped storage unit and irrigation system, and energy trad-
ing cost with main grid. The point estimated method is used
to model the uncertainties of wind power and wholesale elec-
tricity price. A probabilistic scenario-based optimal day-ahead
economic operation of a grid-connected hybrid AC-DC MG is
presented in [46]. It minimizes overall operation cost that in-
cludes energy trading cost with the main grid and operating cost
of CGs. The forecasted values of electricity price, solar power,
wind power, AC and DC loads are used. The scenarios with as-
signed probabilities are considered to deal with uncertainty in
forecasted values of electricity price, solar power, wind power,
AC and DC loads. A proportional-integral controller is used
to control battery current that reduces fluctuations in DC bus
voltage.
Cau et al. [124] presented a stochastic EMS for an iso-
lated MG to minimize the utilization cost of battery and hy-
drogen storage system, and penalty cost on load shedding and
dumped power. The utilization cost of ESS includes operation
and maintenance costs, depreciation cost, and replacement cost.
Scenario tree-based approach models the uncertainties of solar
irradiance, wind speed, and load demand. Scenario reduction
method limits these number of scenarios to reduce processing
time of the proposed approach. For hydrogen storage opera-
tion, the electrolyzer converts the excess energy into hydrogen
and stores it into H2 tanks. These H2 tanks feed a fuel cell
for electricity generation in case of little or no availability of
renewable generation to meet the load demand. The proposed
approach is shown to be more efficient and effective than EMS
with perfect forecast and SOC-based EMS in minimizing oper-
ational cost of MG. A two stage stochastic EMS is presented in
[125] for efficient energy scheduling of a grid-connected MG. It
aims to minimize the operating cost of CGs, battery degradation
cost, and energy trading cost with the main grid. The two-stage
model optimizes the day-ahead operation of MG in a first stage,
and performs AC power flow, as a real-time operation, to com-
pute power losses in a second one. Both these stages are solved
iteratively until net change in power loss is within the specified
limits. The effectiveness of the proposed model is tested on an
IEEE 37-node system.
Rezaei et al. [50] introduced a stochastic frequency secu-
rity constrained EMS for an isolated droop controlled MG. It
minimizes frequency deviations during day-ahead MG opera-
15
Table 9: Critical analysis of MG EMSs based on other artificial intelligent methods.
Ref. Proposed
approach
Contributions Limitations Supervisory
control
Uncertainty
handling
approach
[117] Game theory Objectives are to achieve maximum utilization on pro-
sumers level and maximum profit on MG level. A billing
mechanism is introduced to handle uncertainties of solar
power and load demand.
ESS is ignored, which can cause load shed-
ding during islanded operation in case of dis-
turbance in main grid because DR and PV can-
not meet load demand always.
Decentralized Billing
mechanism
[118] Game theory Objectives are to maximize MG and prosumers profits
individually, while satisfying Stackelberg equilibrium.
Emission cost of microturbine is not consid-
ered.
Decentralized Forecasted
[119] Game theory A bi-level problem is defined to minimize operational
cost of MGs network in upper level and maximize net
profit of each MG in lower level.
Computational complexity is not discussed.
No ESS is considered.
Decentralized Forecasted
[120] Modified game
theory
A multi-objective EMS of a grid-connected MG is pro-
posed to minimize its operational and emission costs.
Computational time complexity of the pro-
posed approach is not discussed. DR is not
considered.
Centralized Forecasted
[121] Markov decision
process
Electricity and natural gas costs are minimized consid-
ering wind power uncertainties to optimize the operation
of MG.
Higher DOD results in fast degradation of bat-
tery. Emission cost of CG is not considered.
Centralized Markov decision
process
[122] Adaptive
intelligence
technique
MG EMS is proposed to maximize utilization of RERs
and minimize load fluctuations by managing operation
of hybrid ESS.
Neither MG profit nor selection of DOD of bat-
tery are discussed.
Centralized Markov chain
Monte Carlo
method
tion and limits operational and emission costs of MG to a rea-
sonable level. Operational cost of MG includes operating cost
of CGs, RERs and reserves, load shedding cost, and DR in-
centives. Scenario generation and reduction method is used to
model the uncertainties of RERs and load demand by using the
probability density functions of their forecasted errors. CGs
outages-based contingency analysis is also studied to test the
robustness of the proposed approach. Shen et al. [126] pre-
sented a scenario-based stochastic EMS for a grid-connected
MG using conditional value at risk methodology that aims to
maximize the expected profit. The conditional value at risk is
used to consider the risk level in computing the expected profit
of MG. Latin hybercube sampling-based Monte Carlo simula-
tion method generates scenarios of RERs, load demand, and
electricity price. To reduce computational time, generated sce-
narios are reduced without the loss of accuracy of results. The
sensitivity analysis of energy trade with the main grid is studied
with respect to price standard deviation, price expected value,
DR, and confidence level.
A two-stage stochastic programming model is presented in
[127] to optimize operation of a grid-connected MG EMS con-
sidering uncertainties of RERs and load demand. First stage is
related to optimization of investment cost of MG. Second stage
deals with the energy management operation of MG. The ef-
fects of battery capacity on purchased power, sold power, and
battery SOC are also studied for three cases, which are 50% un-
certainty in load demand, 50% uncertainty in supply, and 50%
uncertainty in both supply and load demand. Farzin et al. [128]
presented a stochastic EMS for optimal operation of an MG
during unforeseen islanding periods. The scenarios of island-
ing durations and their corresponding probability of occurrence
are estimated based on normal distribution function for each is-
landing event, which is caused by disturbances or failures in
the main grid. Scenario generation and reduction methods is
used to take into account the uncertainties of wind power and
load demand during these estimated islanding intervals. The
objective is to minimize the expected operational cost of MG,
which includes operating cost of microturbine, wind power and
battery, load shedding cost, potential revenue, and risk factor
associated with the objective function value. The sensitivity
analysis is also presented to study the impacts of islanding du-
ration, risk factor, and battery operational cost.
A multi-objective stochastic EMS that considers portable
RERs for DR application is presented in [42] to minimize op-
erational and emission costs of a grid-connected MG. The op-
erational cost of MG includes operating cost of dispatch units,
RERs, ESSs, and energy trading cost. The portable RERs in-
clude small scale wind turbine and photovoltaic system. The
developed multi-objective stochastic optimization model is sol-
ved by augmented epsilon constraint method in GAMS soft-
ware. The impacts of electricity price, combined heat and power
coefficient, initial energy states of ESSs, cut-in and cut-out wind
speed, grid-tie line limit, and minimum reserves on operational
and emission costs of MG are also studied. Liu et al. [129]
proposed a chance constrained programming approach-based
EMS for peak power shaving and frequency regulation of a
grid-connected MG under uncertainties of RERs and load de-
mand. The MG energy management operation is divided into
two sub-problems of energy magnitude scheduling within the
defined energy boundaries for system protection and real-time
energy capacity deviation limit for frequency regulation. The
proposed approach is more efficient than greedy planning, ro-
bust programming, and scenario-based optimization methods in
terms of cost savings.
Kuznetsova et al. [130] proposed a decentralized EMS for a
grid-connected MG using agent-based modeling and robust op-
timization approach to improve its performance and reliability.
MG performance is determined in terms of imbalance cost asso-
ciated with uncertainties of power output of RERs, load demand
and electricity price. Moreover, MG reliability is improved by
considering loss of expected energy and loss of load expecta-
tion parameters. The maximum deviations in power output of
RERs, demand and electricity price forecasting are estimated
by non-dominated Sorting GA trained NN. These maximum
deviation ranges are used in formulating a robust optimization
model for optimal operation of MG. Agent-based modeling is
16
used to achieve the decision making of each agent to minimize
expenses of train station and residential district. It also maxi-
mizes revenue of wind power unit at individual level. A sce-
nario-based robust optimization approach is presented in [131]
to optimize the worst case realization of energy scheduling of
a grid-connected MG. The uncertainty sets of RERs and load
demand are determined by interval prediction theory based on
their forecast errors. Taguchi orthogonal array method uses
these sets to generate limited number of scenarios. Scenarios
with best statistical information are selected to reduce the pro-
cessing time as compared to Monte Carlo simulation. Robust
EMS model is developed to minimize social benefit cost of MG.
It includes operating cost of CG and battery together with en-
ergy trading cost worst realization using search strategy based
on Taguchi orthogonal array method.
Hu et al. [132] introduced a two-stage robust optimization-
based grid-connected MG EMS model. It performs day-ahead
unit commitment operation in a first stage, and real-time eco-
nomic dispatch and energy trade operation in a second stage.
The complexity of the proposed model is relaxed by using Lya-
punov optimization method. The efficacy and efficiency of the
proposed approach is better than greedy algorithm. Zhang et
al. [133] introduced a robust decentralized EMS for a grid-
connected MG to minimize its social cost. The objective func-
tion includes operational cost of CGs and battery, DR incentives
and worst case energy trading cost. The developed optimization
model is decomposed and solved iteratively by a distributed im-
plementation of subgradient method. The authors also analyzed
that the social cost of MG decreases with increase in proportion
of selling energy price to buying energy price.
An optimal EMS of a grid-connected MG is presented in
[134]. The nonlinear thermal and electric efficiency curve of
microturbine is converted into piecewise linear approximations
to make the problem tractable. The objective is to minimize
the operational cost of MG that includes energy trading cost
with the main grid, operating cost of combined cooling, heat
and power unit, and battery degradation cost. The impacts of
different budget of uncertainty values and efficiency curves on
MG performance are also studied. Two-stage robust optimiza-
tion approach for EMS of a grid-connected MG is presented in
[51]. It performs day-ahead unit commitment operation of CGs
in a first stage, and real-time DR and energy trading operation
in a second stage. The proposed model takes into account the
uncertainties of RERs and MG islanding events. The budget of
uncertainty factor is introduced to avoid over-conservatism of
the proposed approach, resulting in tradeoffbetween optimality
and robustness. The developed two-stage model is transformed
into a large scale MILP problem, which is solved by column
and constraint generation algorithm-based decomposition ap-
proach. The comparison with stochastic programming method
shows that the proposed approach is more effective in finding
worst-case best solution. However, it is outperformed with re-
spect to average case best solution.
The critical analysis of MG EMSs based on stochastic and
robust programming approaches is summarized in Table 10.
The main limitations of these approaches are complex problem
formulation, higher computational time complexity, no mitiga-
tion mechanism for reducing GHG emissions, higher DOD se-
lection of battery, and customer privacy issues.
4.5. EMS based on Model Predictive Control
Solanki et al. [135] emphasized on use of smart load to per-
form an efficient EMS strategy based on a model predictive con-
trol (MPC) approach for an islanded MG. A supervised NN is
used to estimate the residential controllable load. The objective
function consists of operating cost of CGs and penalty cost on
energy curtailment. The developed optimization model shows
improved performance than decoupled EMS model, which de-
composes the main problem into two sub-problems, namely:
unit commitment and optimal power flow. It performs better in
terms of best solution, less energy curtailed, reduced peak de-
mand and improved load factor, but at the cost of higher com-
putational time. The impacts of DR on objective function, less
energy curtailment, load factor improvement, and peak demand
reduction are also studied. Mendes et al. [136] proposed a hi-
erarchical structure for optimal energy management of a grid-
connected MG, together with vehicle-to-grid operation, to en-
sure its stability in a first level and its economic operation in a
second one. The economic operation of MG includes energy
trade with main grid, maximum use of RERs, and management
of battery and EVs operation. The effectiveness of the proposed
approach is also validated experimentally.
Minchala-Avila et al. [137] presented a nonlinear MPC ap-
proach for an islanded MG EMS model to ensure its stable op-
eration. The voltage stability of MG system is secured by man-
aging the control of battery SOC and load shedding scheme. An
artificial NN is used for load demand prediction and an adaptive
neuro-fuzzy inference system is used to forecast power output
of CG based on load demand. Both of these predication meth-
ods are tested with the proposed approach and the latter shows
better performance. An MPC-based EMS of a grid-connected
MG is presented in [138] to minimize the energy trading cost
with the main grid, and to ensure better utilization of battery
during peak load demand. It also ensures maximum use of wind
power to meet local demand. A fault-tolerant control scheme
is introduced to have smooth operation of MG during sudden
failures of wind turbines and power supply shortage. The ef-
ficacy and efficiency of the proposed approach is better than
reinforcement learning algorithm in respect of best solution. A
robust EMS for an isolated MG, located in a Chilean village, is
proposed in [139]. The fuzzy prediction interval method deter-
mines the uncertainty set for wind energy. The uncertainty of
solar power is not considered due to the low probability of solar
irradiance variability. The objective is the efficient dispatch of
MG sources at minimum operating cost. Two MPC optimizers
are used to determine the upper and lower range of dispatch of
MG units and the final output is obtained by a convex sum of
these ranges with a defined weighting factor.
Luo et al. [140] introduce a two-stage coordinated approach
for efficient EMS operation of a grid-connected MG. The piece-
wise linear approximations of thermal and electric efficiency
curves of microturbine are considered in a first stage. It in-
cludes the operating cost of microturbine, battery degradation
cost, and energy trading cost with the main grid in objective
17
Table 10: Critical analysis of MG EMSs based on stochastic and robust programming methods.
Ref. Proposed
approach
Contributions Limitations Supervisory
control
Uncertainty
handling
approach
[123] Stochastic
optimization
The overall cost of MG is minimized, including opera-
tional cost of pumped storage unit and irrigation system,
and energy trading cost.
The levelized cost of energy of wind and power
losses cost are ignored. DR is not considered.
Centralized Two point
estimate method
[46] Scenario-based
optimization
The overall cost of hybrid AC-DC MG is minimized,
including energy trading cost with main grid and oper-
ating cost of CGs. A battery current controller is used to
regulate the fluctuations in DC bus voltage.
Emission cost of dispatch units and DR are not
taken into consideration.
Centralized Scenario
generation method
[124] Stochastic
optimization
The utilization cost of battery and hydrogen storage sys-
tem, and penalty cost on load shedding and dumped
power are minimized for efficient operation of MG.
Computational time complexity is higher. Centralized Scenario tree
approach
[125] Stochastic
programming
The operating cost of CGs, battery degradation cost and
energy trading cost with the main grid are minimized.
Computational time complexity is not dis-
cussed. Higher DOD results in fast degradation
of battery.
Centralized Monte Carlo
simulation method
[50] Stochastic
optimization
The frequency management-based EMS is defined to
control frequency deviations of an isolated MG while
satisfying techno-economic and environmental con-
straints.
Computational time complexity of proposed
approach is not discussed.
Centralized Monte Carlo
simulation and
scenario reduction
approaches
[126] Stochastic
programming
The expected profit of MG is maximized considering
risk management based on conditional value at risk fac-
tor.
The levelized cost of energy of RERs and bat-
tery degradation cost are not considered.
Centralized Monte Carlo
simulation and
scenario reduction
[127] Stochastic
programming
Two-stage stochastic framework for MG EMS is pro-
posed to minimize investment cost in first stage and op-
eration cost in second stage.
Emission cost of gas turbine and battery degra-
dation cost are not considered.
Centralized Scenario tree
approach
[128] Stochastic
optimization
Expected operational cost of MG is minimized for its
optimal operation during unexpected islanding events.
Computational time complexity is not dis-
cussed. DR is also not taken into considera-
tion.
Centralized Monte Carlo
simulation and
backward
reduction methods
[42] Stochastic
optimization
Multi-objective EMS model is proposed to minimize op-
erational and emission costs of MG.
Selection of higher DOD results in fast degra-
dation of battery.
Centralized Scenario
generation and
reduction
[129] Chance
constrained
programming
Two EMS models are represented for peak shaving and
system protection, and frequency regulation, respec-
tively.
Complex formulation. Tradeoffbetween com-
plexity and cost is ignored.
Centralized Probabilistic
constraints
[130] Robust
optimization
MG performance in terms of imbalance cost and MG
reliability in terms of loss of expected energy and loss
of load expectation are realized.
Complex formulation. DR in residential dis-
trict is ignored.
Decentralized Bounded
uncertainty
[131] Robust
optimization
Social benefit cost including operating cost of CG and
battery, and energy trading cost is minimized while re-
alizing worst case scenario for energy trading cost to ac-
count for the MG uncertainties.
Emission cost of CG and DR are not taken into
consideration. Higher DOD is selected that
leads to fast degradation of battery.
Centralized Bounded
uncertainty
[132] Robust
optimization
Two-stage EMS is proposed to perform day-ahead unit
commitment operation in a first stage and real-time eco-
nomic dispatch and energy trading operation in a second
one.
Computational time complexity of proposed
approach is not discussed. Higher DOD de-
creases battery lifetime exponentially.
Centralized Bounded
uncertainty
[133] Robust
optimization
Social cost of MG is minimized that includes operating
cost of CGs and battery, DR incentives and worst case
energy trading cost.
Complex formulation. Computation time com-
plexity of proposed method is also not dis-
cussed.
Decentralized Bounded
uncertainty
[134] Robust
optimization
Operational cost of MG, which includes energy trad-
ing cost with the main grid, operating cost of combined
cooling, heat and power unit and battery degradation
cost, is minimized.
DR application in handling uncertainties of
MG system is not considered.
Centralized Bounded
uncertainty
[51] Robust
optimization
Two-stage EMS model performs day-ahead unit com-
mitment operation of CGs in first stage and real-time
DR and energy trading operation in second stage con-
sidering MG islanding events.
Levelized cost of energy of RERs is not con-
sidered. Computation time complexity of pro-
posed method is also not discussed.
Centralized Bounded
uncertainty
function. The first stage is related to economic dispatch oper-
ation of MG, while the second stage is an adjustment one. It
deals in real-time with any load generation imbalance caused
by uncertainties of power output of RERs and load demand. A
sustainable EMS for an islanded MG is presented in [141] to op-
timize its energy scheduling considering multi-objective func-
tions of its operation and emission costs. Different operational
strategies have been considered with respect to minimization
of operation and emission costs of MG. The impacts of DR on
reduction in operation and emission costs are also studied.
Prodan et al. [142] proposed a fault-tolerant EMS for a grid-
connected MG considering generator faults and uncertainties
of RERs, load demand, and market price. The objective func-
tion includes the battery lifecycle cost, external grid cost, and
load generation imbalance cost. The battery lifecycle cost is
included in the objective function to consider its aging factor.
Finally, a fault-tolerant strategy is introduced as a constraint to
ensure the availability of sufficient resources of ESS to meet the
load demand during generators fault.
Table 11 summarizes the contributions and main limitations
18
of the proposed MG EMSs based on model predictive control.
Centralized supervisory control architecture is used in all these
MG EMSs. Moreover, forecasted data are mainly used to take
into account uncertainties in these EMSs. However, a lot of
work is still required to focus on computation time complexity,
reduction of GHG emissions of CGs, outages and losses costs
of MG system, efficient integration of DR and EVs, and cus-
tomer privacy issues.
4.6. EMSs based on other approaches
Guo et al. [143] presented a real-time EMS of a standalone
MG that uses rolling horizon optimization method to minimize
the consumption of CG and to maximize utilization of wind en-
ergy. The GA-based back propagation NN method is used for
an hour-ahead wind speed prediction. The resource scheduling
of MG is divided into an hour-ahead scheduling and real-time
dispatch. For real-time dispatch, the two operation modes are
considered. In first mode, battery is used as a main source and
V/fmethod controls its operation. While in second mode, CG
is considered as a main source. Therefore, V/fand PQ methods
are used to control the operations of CG and battery, respec-
tively. In both of these modes, MPPT method is used for wind
turbine. In [144], a two-stage hierarchical method based EMS
is presented to minimize MG operational cost and fluctuations
in energy exchanged at PCC. The proposed method consists of
a day-ahead economic dispatch stage and a two-layer intra-hour
adjustment stage. The latter stage is used to minimize the fluc-
tuations in energy exchanged at PCC in both short term (15
min) and ultra-short term (1 min) intervals. Office building,
modeled as a thermal ESS, and EVs are considered as flexible
resources.
Akter et al. [145] proposed a hierarchical model for power
sharing among residential prosumers in a grid-connected resi-
dential MG. Each house is equipped with a central EMS con-
troller to share information among residential prosumers. Three
types of houses are considered, namely: traditional, proactive,
and enthusiastic. Photovoltaic system is installed in both proac-
tive and enthusiastic houses, while battery is installed in enthu-
siastic houses only. The objective is to maximize the energy
sharing among houses in comparison with the main grid and
minimize the investment cost. The authors concluded that the
payback period is reduced due to power sharing among resi-
dential prosumers. Wang et al. [52] introduced a hierarchal
control for a standalone MG EMS to improve the economi-
cal performance and reliability of MG. The economical objec-
tive includes penalty cost on starting new CGs, penalty cost on
heavy-load and light-load operation of CGs, and operating cost
of CGs and ESS. For system reliability, the battery SOC is used
for voltage and frequency regulation to take into account uncer-
tainties of wind speed and load demand in real-time.
Yanine et al. [146] presented homeostatic control-based en-
ergy and exergy management of a grid-connected MG that aims
to achieve energy efficiency and thriftiness objectives. Home-
ostatic is a biologic term that refers to the process towards the
state of equilibrium. Li-ion battery is used as an energy buffer
to ensure effective implementation of MG demand response op-
eration. In [147], the objective is to utilize the benefits of DR
to deal with power mismatches that are caused by uncertain-
ties of RERs. Predefined set of criteria are used to enhance the
homeostatic regulation and control in energy consumption for
sustainable operation of MG.
Shi et al. [148] proposed a predictor corrector proximal
multiplier algorithm-based distributed EMS for a grid-connect-
ed MG to optimize its operation and to ensure customer privacy.
The objective function includes operational cost of CGs, battery
degradation cost, energy trading cost with main grid, load shed-
ding cost, and power losses. The weights are introduced with
each objective to have a tradeoffbetween operational cost and
network losses minimizations of MG system. The nonconvex-
ity of the developed model is solved by converting it into a re-
laxed optimal power flow problem. The authors concluded that
the proposed approach is more efficient in minimizing network
loss and operation cost of MG as compared to other existing
methods in the literature.
Mohamed and Koiva [149] proposed an online EMS for a
grid-connected MG, based on a mesh adaptive direct search
method, to minimize its overall cost. The overall cost of MG
includes operating cost of CGs, energy trading cost with main
grid, and emission cost. In emission cost, the costs of nitrogen
oxides, carbon oxides, and sulfur oxides are considered. The
proposed approach is concluded to be more efficient than se-
quential quadratic programming approach in minimizing total
cost of MG.
Table 12 summarizes the comparative analysis of the pro-
posed MG EMS models based on other approaches such as
rolling horizon, hierarchical control, homeostatic control, and
many other etc. In these EMSs and solution approaches, an ex-
tensive work is required to focus on computational time com-
plexity, algorithm convergence to optimal solution, reduction
of GHG emission, integration of DR, MG reliability during is-
landed operation, minimization of power losses of MG system,
and customer privacy issues.
5. Real World Applications and Discussion
An EMS is very important in utility, industry, commercial
and residential sectors for energy efficient operation. It aims
to optimize DERs scheduling, reduce energy consumption, and
minimize GHG emissions. The integration of EMS with super-
visory, control and data acquisition (SCADA), and human-to-
machine interface (HMI) helps it in monitoring and analyzing
data. It includes power output of generation sources, weather
forecast, load demand, and real-time energy price. The EMS
uses this data to optimize the system performance at genera-
tion, transmission, and distribution ends.
Most of the MG EMSs studied in the literature, as described
in section 4, have centralized supervisory control architecture.
However, due to the increased penetration of DERs in power
system, centralized architecture faces problems of high compu-
tational time, less system scalability, and high instability in case
of failures. Therefore, researchers are recently more focusing
on a decentralized supervisory control architecture. However,
it requires continuous availability of two-way communication
19
Table 11: Critical analysis of MG EMSs based on model predictive control.
Ref. Proposed
approach
Contributions Limitations Supervisory
control
Uncertainty
handling
approach
[135] Model predictive
control
Operating cost of CGs and penalty cost on energy cur-
tailment are minimized. Moreover, effects of DR on MG
performance are also studied.
Computation time of the proposed approach is
higher. Emission cost of CGs is ignored.
Centralized Artificial NN
[136] Model predictive
control
An EMS is proposed to maintain MG stability in a first
level and to optimize its economic operation in a second
one.
Fully discharged EVs result in exponential de-
crease of their batteries lifetime.
Centralized Forecasted
[137] Model predictive
control
The voltage stability of MG is achieved by controlling
battery SOC and implementing load shedding scheme.
Computational time complexity of proposed
approach is not discussed.
Centralized Forecasted
[138] Model predictive
control
The energy trading cost with main grid is minimized
as well as the maximum utilization of battery and wind
power are achieved.
Operational cost of wind turbine and battery
are not taken into consideration.
Centralized Forecasted
[139] Model predictive
control
Operating cost of CG and penalty cost on load shedding
are minimized. Wind power uncertainty is determined
by fuzzy prediction interval method.
Emission cost of CG and levelized cost of en-
ergy of wind turbine are not considered.
Centralized Fuzzy prediction
interval method
[140] Model predictive
control
Two-stage EMS model is developed which performs
economical scheduling operation in first stage and ad-
justment operation in second stage to deal with uncer-
tainties of power output of RERs and load demand.
Computational time complexity of proposed
approach is not discussed. Emission cost of
CG is not considered.
Centralized Forecasted
[141] Model predictive
control
The multi-objective function of operational cost and
emission cost of MG system are minimized. Impacts
of DR on proposed approach are also studied.
Operational cost of battery is not taken into
consideration.
Centralized Artificial NN
[142] Model predictive
control
Operational cost of MG, which includes battery lifecycle
cost, energy trading cost with main grid and load gener-
ation imbalance cost, are minimized considering gener-
ator faults and uncertainties.
Levelized cost of energy of RERs is not con-
sidered.
Centralized Bounded
uncertainty
Table 12: Critical analysis of MG EMSs based on other methods.
Ref. Proposed
approach
Contributions Limitations Supervisory
control
Uncertainty
handling
approach
[143] Rolling horizon
optimization
Minimum consumption of DG and maximum utilization
of wind energy are achieved. V/f,PQ and MPPT meth-
ods are used for real-time operation and control of CG
and wind turbine.
Operational cost of MG is not considered.
Power losses of proposed MG system are also
ignored.
Centralized GA-based back
propagation NN
[144] Hierarchical
control
Operational cost of MG is minimized in first stage, while
second stage is used to reduce fluctuations in energy ex-
changed at PCC.
Computational time complexity of proposed
approach is not discussed.
Centralized Forecasted
[145] Hierarchical
control
Objectives of minimization of investment cost of MG
and maximization of energy sharing among houses are
achieved.
Battery operation is not discussed. Comfort
level of users is also not considered.
Centralized Forecasted
[52] Hierarchical
control
Economical operation and reliability of MG is realized
by minimization of operational cost, and regulation of
system voltage and frequency, respectively.
Levelized cost of energy of RERs is not con-
sidered.
Centralized Forecasted
[146] Homeostatic
control
Demand response, energy efficiency and thriftiness ob-
jectives are achieved in energy and exergy management
of a grid-connected MG.
Battery degradation cost is not taken into con-
sideration.
Centralized Forecasted
[147] Homeostatic
control
The energy efficiency and thriftiness objectives are
achieved by effective implementation of demand re-
sponse in the residential sector.
The benefits of DR in terms of mitigating GHG
emissions and system reliability can also be re-
alized.
Centralized Forecasted
[148] Predictor
corrector proximal
multiplier
algorithm
Optimal operation of MG is achieved by minimizing op-
erational cost of CGs, battery degradation cost, energy
trading cost with main grid, load shedding cost, and
power losses.
Computational time complexity of proposed
approach is not discussed.
Decentralized Forecasted
[149] Mesh adaptive
direct search
Total cost of MG, which includes operating cost of CGs,
energy trading cost and emission cost, is minimized.
Levelized cost of energy of RERs and battery
degradation cost are not taken into considera-
tion.
Centralized Forecasted
link among MG components and their synchronization that re-
sults in increased system cost. Moreover, upgradation cost of
these communication systems require to be optimized.
The operation of MG EMS is divided into two layers, day-
ahead dispatch layer and real-time dispatch layer. The day-
ahead energy dispatch is further divided into sub-hourly dis-
patch to take into account the forecast errors. The reference
values are sent to LCs in real-time using communication links.
For rural, residential, and remote areas MGs, the selection of
communication technologies mainly depends on the deploy-
ment cost and data rate. Zigbee, z-wave, wifi, and Bluetooth
are better options for such MGs. However for municipality and
utility microgrids, the coverage range and data rate are most
important, and passive optical network, 3G and 4G are better
options. At DERs and load end, routers use these communica-
tion technologies to share information among LCs and MGCC.
20
The LCs can be implemented by using low cost embedded sys-
tems, such as Aurdinos and Rasberry PI. They are designed to
collect data from monitoring sensors and smart meters and per-
form control actions locally to ensure customer privacy. The
MGCC performs energy management operations with the help
of SCADA, HMI and information received from LCs. The so-
lution approaches for these energy management operations are
mainly selected on computational time complexity and conver-
gence to optimum solution based merits. The experimental im-
plementation of microgrid energy management systems are also
validated using various solution approaches such as linear pro-
gramming [66, 71], meta-heuristic methods [93, 94, 97, 103],
artificial intelligent [105], and model predictive control [136].
Moreover, microgrid energy management systems are currently
being developed and deployed by energy companies as Schnei-
der Electric[150], ABB [151], General Electric [152], Siemens
[153], Alstom, Tesla, and so forth.
6. Conclusion and Future Trends
Microgrids are generally composed of distributed energy re-
sources, demand response, electric vehicles, local controllers,
microgrid energy management system-based central controller,
and communication devices. This paper has presented a com-
prehensive and critical review on the developed microgrid en-
ergy management strategies and solution approaches. The main
objectives of the energy management system are to optimize the
operation, energy scheduling, and system reliability in both is-
landed and grid-connected microgrids for sustainable develop-
ment. Hence, microgrid energy management system is a multi-
objective topic that deals with technical, economical, and envi-
ronmental issues.
This extensive critical review addresses solutions, opportu-
nities, and prospects to achieve the energy management objec-
tives using various efficient methods. These methods are se-
lected based on their suitability, practicability, and tractability,
for optimal operation of microgrids. The objective types of MG
EMS depend on its operation mode, its centralized or decen-
tralized operation, economical aspects, and the intermittent and
volatile nature of renewable energy sources. They also consider
environmental issues of conventional generators, health status
of batteries, active DR integration, system losses and reliabil-
ity, and customer privacy. Many research studies have been
conducted on some of these objective types. However, an ex-
tensive is still required to manage the customer privacy issues,
secure and reliable communication system cost management,
particularly for decentralized operation. Furthermore, micro-
grid systems reliability analysis is not studied in detail for is-
landing and remote applications. These potential areas need to
be addressed in detail to achieve optimal energy efficient oper-
ation of microgrids.
References
[1] J. Wu, J. Yan, H. Jia, N. Hatziargyriou, N. Djilali, H. Sun, Integrated
Energy Systems, Applied Energy 167 (2016) 155–157.
[2] J. Keirstead, M. Jennings, A. Sivakumar, A review of urban energy sys-
tem models: Approaches, challenges and opportunities, Renewable and
Sustainable Energy Reviews 16 (6) (2012) 3847–3866.
[3] H. Hu, N. Xie, D. Fang, X. Zhang, The role of renewable energy con-
sumption and commercial services trade in carbon dioxide reduction:
Evidence from 25 developing countries, Applied Energy 211 (2018)
1229 – 1244.
[4] S. Parhizi, H. Lotfi, A. Khodaei, S. Bahramirad, State of the Art in Re-
search on Microgrids: A Review, IEEE Access 3 (2015) 890–925.
[5] S. Bacha, D. Picault, B. Burger, I. Etxeberria-Otadui, J. Martins, Pho-
tovoltaics in Microgrids: An Overview of Grid Integration and En-
ergy Management Aspects, IEEE Industrial Electronics Magazine 9 (1)
(2015) 33–46.
[6] N. Hatziargyriou, H. Asano, R. Iravani, C. Marnay, Microgrids 5 (4)
(2007) 78–94.
[7] DOE, Summary report: 2012 DOE microgrid workshop, Chicago, Illi-
nois, 2012.
[8] R. Lasseter, A. Akhil, C. Marnay, J. Stevens, J. Dagle, R. Guttromson,
S. A. Meliopoulous, R. Yinger, J. Eto, The certs microgrid concept,
white paper on integration of distributed energy resources, California
Energy Commission, Office of Power Technologies-US Department of
Energy, LBNL-50829, 2002.
[9] REN21, Renewables 2017-global status report, Paris, REN21 Secretari-
ate, 2017.
[10] N. Hatziargyriou, et al., Microgrids: architectures and control, Wiley
Online Library, 2014.
[11] A. H. Fathima, K. Palanisamy, Optimization in microgrids with hybrid
energy systems–a review, Renewable and Sustainable Energy Reviews
45 (2015) 431–446.
[12] A. A. Khan, M. Naeem, M. Iqbal, S. Qaisar, A. Anpalagan, A com-
pendium of optimization objectives, constraints, tools and algorithms for
energy management in microgrids, Renewable and Sustainable Energy
Reviews 58 (2016) 1664–1683.
[13] L. Meng, E. R. Sanseverino, A. Luna, T. Dragicevic, J. C. Vasquez,
J. M. Guerrero, Microgrid supervisory controllers and energy manage-
ment systems: A literature review, Renewable and Sustainable Energy
Reviews 60 (2016) 1263–1273.
[14] V. N. Coelho, M. W. Cohen, I. M. Coelho, N. Liu, F. G. Guimar£es,
Multi-agent systems applied for energy systems integration: State-of-
the-art applications and trends in microgrids, Applied Energy 187 (2017)
820 – 832.
[15] L. Olatomiwa, S. Mekhilef, M. Ismail, M. Moghavvemi, Energy man-
agement strategies in hybrid renewable energy systems: A review, Re-
newable and Sustainable Energy Reviews 62 (2016) 821 – 835.
[16] C. Gamarra, J. M. Guerrero, Computational optimization techniques ap-
plied to microgrids planning: a review, Renewable and Sustainable En-
ergy Reviews 48 (2015) 413–424.
[17] N. Lidula, A. Rajapakse, Microgrids research: A review of experimen-
tal microgrids and test systems, Renewable and Sustainable Energy Re-
views 15 (1) (2011) 186–202.
[18] S. A. Gopalan, V. Sreeram, H. H. Iu, A review of coordination strate-
gies and protection schemes for microgrids, Renewable and Sustainable
Energy Reviews 32 (2014) 222–228.
[19] M. Gayatri, A. M. Parimi, A. P. Kumar, A review of reactive power com-
pensation techniques in microgrids, Renewable and Sustainable Energy
Reviews 81 (2018) 1030–1036.
[20] M. H. Andishgar, E. Gholipour, R.-a. Hooshmand, An overview of con-
trol approaches of inverter-based microgrids in islanding mode of opera-
tion, Renewable and Sustainable Energy Reviews 80 (2017) 1043–1060.
[21] U. B. Tayab, M. A. B. Roslan, L. J. Hwai, M. Kashif, A review of droop
control techniques for microgrid, Renewable and Sustainable Energy
Reviews 76 (2017) 717–727.
[22] S. M. Malik, X. Ai, Y. Sun, C. Zhengqi, Z. Shupeng, Voltage and fre-
quency control strategies of hybrid ac/dc microgrid: a review, IET Gen-
eration, Transmission & Distribution 11 (2) (2017) 303–313.
[23] B. M. Eid, N. A. Rahim, J. Selvaraj, A. H. El Khateb, Control methods
and objectives for electronically coupled distributed energy resources in
microgrids: A review, IEEE Systems Journal 10 (2) (2016) 446–458.
[24] S. M. Dawoud, X. Lin, M. I. Okba, Hybrid renewable microgrid opti-
mization techniques: A review, Renewable and Sustainable Energy Re-
views 82 (2018) 2039 – 2052.
21
[25] J. J. Justo, F. Mwasilu, J. Lee, J.-W. Jung, Ac-microgrids versus dc-
microgrids with distributed energy resources: A review, Renewable and
Sustainable Energy Reviews 24 (2013) 387–405.
[26] F. F. Yanine, E. E. Sauma, Review of grid-tie micro-generation systems
without energy storage: Towards a new approach to sustainable hybrid
energy systems linked to energy efficiency, Renewable and Sustainable
Energy Reviews 26 (2013) 60 – 95.
[27] K. P. Kumar, B. Saravanan, Recent techniques to model uncertain-
ties in power generation from renewable energy sources and loads in
microgrids–a review, Renewable and Sustainable Energy Reviews 71
(2017) 348–358.
[28] J. Pan, R. Jain, S. Paul, A survey of energy efficiency in buildings and
microgrids using networking technologies, IEEE Communications Sur-
veys & Tutorials 16 (3) (2014) 1709–1731.
[29] M. H. Amrollahi, S. M. T. Bathaee, Techno-economic optimization of
hybrid photovoltaic/wind generation together with energy storage sys-
tem in a stand-alone micro-grid subjected to demand response, Applied
Energy 202 (2017) 66–77.
[30] F. F. Yanine, F. I. Caballero, E. E. Sauma, F. M. Crdova, Homeostatic
control, smart metering and efficient energy supply and consumption
criteria: A means to building more sustainable hybrid micro-generation
systems, Renewable and Sustainable Energy Reviews 38 (2014) 235 –
258.
[31] M. Abedini, M. H. Moradi, S. M. Hosseinian, Optimal management of
microgrids including renewable energy scources using gpso-gm algo-
rithm, Renewable Energy 90 (2016) 430–439.
[32] F. Caballero, E. Sauma, F. Yanine, Business optimal design of a grid-
connected hybrid pv (photovoltaic)-wind energy system without energy
storage for an easter island’s block, Energy 61 (2013) 248 – 261.
[33] F.-F. Li, J. Qiu, Multi-objective optimization for integrated hydrophoto-
voltaic power system, Applied Energy 167 (2016) 377 – 384.
[34] A. Askarzadeh, A memory-based genetic algorithm for optimization of
power generation in a microgrid, IEEE Transactions on Sustainable En-
ergy PP (99) (2017) 1.
[35] S. Sukumar, H. Mokhlis, S. Mekhilef, K. Naidu, M. Karimi, Mix-mode
energy management strategy and battery sizing for economic operation
of grid-tied microgrid, Energy 118 (2017) 1322–1333.
[36] J. Almada, R. Le ˜
ao, R. Sampaio, G. Barroso, A centralized and heuristic
approach for energy management of an ac microgrid, Renewable and
Sustainable Energy Reviews 60 (2016) 1396–1404.
[37] L. K. Panwar, S. R. Konda, A. Verma, B. K. Panigrahi, R. Kumar, Op-
eration window constrained strategic energy management of microgrid
with electric vehicle and distributed resources, IET Generation, Trans-
mission & Distribution 11 (3) (2017) 615–626.
[38] S. Obara, M. Kawai, O. Kawae, Y. Morizane, Operational planning of
an independent microgrid containing tidal power generators, sofcs, and
photovoltaics, Applied Energy 102 (2013) 1343 – 1357.
[39] J. P. Fossati, A. Galarza, A. Mart´
ın-Villate, J. M. Echeverr´
ıa, L. Font´
an,
Optimal scheduling of a microgrid with a fuzzy logic controlled storage
system, International Journal of Electrical Power & Energy Systems 68
(2015) 61–70.
[40] A. Baziar, A. Kavousi-Fard, Considering uncertainty in the optimal en-
ergy management of renewable micro-grids including storage devices,
Renewable Energy 59 (2013) 158–166.
[41] T. Bogaraj, J. Kanakaraj, Intelligent energy management control for in-
dependent microgrid, S¯
adhan¯
a 41 (7) (2016) 755–769.
[42] V. S. Tabar, M. A. Jirdehi, R. Hemmati, Energy management in micro-
grid based on the multi objective stochastic programming incorporating
portable renewable energy resource as demand response option, Energy
118 (2017) 827–839.
[43] M. Motevasel, A. R. Seifi, T. Niknam, Multi-objective energy manage-
ment of chp (combined heat and power)-based micro-grid, Energy 51
(2013) 123–136.
[44] Nishrina, A. G. Abdullah, A. Risdiyanto, A. Nandiyanto, Hybrid energy
system design of micro hydro-pv-biogas based micro-grid, IOP Confer-
ence Series: Materials Science and Engineering 180 (1) (2017) 012080.
[45] S. A. Arefifar, M. Ordonez, Y. A.-R. I. Mohamed, Energy management
in multi-microgrid systemsdevelopment and assessment, IEEE Transac-
tions on Power Systems 32 (2) (2017) 910–922.
[46] P. T. Baboli, M. Shahparasti, M. P. Moghaddam, M. R. Haghifam,
M. Mohamadian, Energy management and operation modelling of hy-
brid ac–dc microgrid, IET Generation, Transmission & Distribution
8 (10) (2014) 1700–1711.
[47] D. Tenfen, E. C. Finardi, A mixed integer linear programming model for
the energy management problem of microgrids, Electric Power Systems
Research 122 (2015) 19–28.
[48] A. A. Moghaddam, A. Seifi, T. Niknam, M. R. A. Pahlavani, Multi-
objective operation management of a renewable mg (micro-grid) with
back-up micro-turbine/fuel cell/battery hybrid power source, Energy
36 (11) (2011) 6490–6507.
[49] A. Chaouachi, R. M. Kamel, R. Andoulsi, K. Nagasaka, Multiobjective
intelligent energy management for a microgrid, IEEE Transactions on
Industrial Electronics 60 (4) (2013) 1688–1699.
[50] N. Rezaei, M. Kalantar, Stochastic frequency-security constrained en-
ergy and reserve management of an inverter interfaced islanded micro-
grid considering demand response programs, International Journal of
Electrical Power & Energy Systems 69 (2015) 273–286.
[51] Y. Guo, C. Zhao, Islanding-aware robust energy management for micro-
grids, IEEE Transactions on Smart Grid 9 (2) (2016) 1301–1309.
[52] C. Wang, Y. Liu, X. Li, L. Guo, L. Qiao, H. Lu, Energy management
system for stand-alone diesel-wind-biomass microgrid with energy stor-
age system, Energy 97 (2016) 90–104.
[53] B. Bhandari, K.-T. Lee, C. S. Lee, C.-K. Song, R. K. Maskey, S.-H. Ahn,
A novel off-grid hybrid power system comprised of solar photovoltaic,
wind, and hydro energy sources, Applied Energy 133 (2014) 236 – 242.
[54] M. Javidsharifi, T. Niknam, J. Aghaei, G. Mokryani, Multi-objective
short-term scheduling of a renewable-based microgrid in the presence
of tidal resources and storage devices, Applied Energy 216 (2018) 367
– 381.
[55] G. Davis, Integration of distributed energy resources the certs microgrid
concept, CERTS, 2003.
[56] Y. Li, F. Nejabatkhah, Overview of control, integration and energy man-
agement of microgrids, Journal of Modern Power Systems and Clean
Energy 3 (2) (2014) 212–222.
[57] N. Hatziargyriou, N. Jenkins, G. Strbac, J. P. Lopes, J. Ruela, A. En-
gler, J. Oyarzabal, G. Kariniotakis, A. Amorim, et al., Microgrids–large
scale integration of microgeneration to low voltage grids, CIGRE C6-
309, 2006.
[58] M. Li, X. Zhang, G. Li, C. Jiang, A feasibility study of microgrids for
reducing energy use and ghg emissions in an industrial application, Ap-
plied Energy 176 (2016) 138 – 148.
[59] V. C. Gungor, D. Sahin, T. Kocak, S. Ergut, C. Buccella, C. Cecati,
G. P. Hancke, Smart grid technologies: Communication technologies
and standards, IEEE transactions on Industrial informatics 7 (4) (2011)
529–539.
[60] B. Wang, M. Sechilariu, F. Locment, Intelligent dc microgrid with smart
grid communications: Control strategy consideration and design, IEEE
transactions on smart grid 3 (4) (2012) 2148–2156.
[61] A. Usman, S. H. Shami, Evolution of communication technologies for
smart grid applications, Renewable and Sustainable Energy Reviews 19
(2013) 191–199.
[62] F. G´
omez-Cuba, R. Asorey-Cacheda, F. J. Gonzalez-Castano, Smart
grid last-mile communications model and its application to the study of
leased broadband wired-access, IEEE Transactions on Smart Grid 4 (1)
(2013) 5–12.
[63] Y. Kabalci, A survey on smart metering and smart grid communication,
Renewable and Sustainable Energy Reviews 57 (2016) 302–318.
[64] IEC 61970: Energy management system application program interface
(EMS-API), IEC, 2005.
[65] C. Chen, S. Duan, T. Cai, B. Liu, G. Hu, Smart energy management
system for optimal microgrid economic operation, IET renewable power
generation 5 (3) (2011) 258–267.
[66] N. Anglani, G. Oriti, M. Colombini, Optimized energy management sys-
tem to reduce fuel consumption in remote military microgrids, IEEE
Transactions on Industry Applications 53 (6) (2017) 5777–5785.
[67] G. Comodi, A. Giantomassi, M. Severini, S. Squartini, F. Ferracuti,
A. Fonti, D. N. Cesarini, M. Morodo, F. Polonara, Multi-apartment res-
idential microgrid with electrical and thermal storage devices: Exper-
imental analysis and simulation of energy management strategies, Ap-
plied Energy 137 (2015) 854–866.
[68] C. Corchero, M. Cruz-Zambrano, F.-J. Heredia, et al., Optimal energy
management for a residential microgrid including a vehicle-to-grid sys-
22
tem, IEEE transactions on smart grid 5 (4) (2014) 2163–2172.
[69] J. Shen, C. Jiang, Y. Liu, J. Qian, A microgrid energy management
system with demand response for providing grid peak shaving, Electric
Power Components and Systems 44 (8) (2016) 843–852.
[70] P. P. Vergara, J. C. L ´
opez, L. C. da Silva, M. J. Rider, Security-
constrained optimal energy management system for three-phase residen-
tial microgrids, Electric Power Systems Research 146 (2017) 371–382.
[71] A. C. Luna, L. Meng, N. L. Diaz, M. Graells, J. C. Vasquez, J. M. Guer-
rero, Online energy management systems for microgrids: Experimen-
tal validation and assessment framework, IEEE Transactions on Power
Electronics 33 (3) (2018) 2201–2215.
[72] D. E. Olivares, C. A. Ca˜
nizares, M. Kazerani, A centralized energy man-
agement system for isolated microgrids, IEEE Transactions on smart
grid 5 (4) (2014) 1864–1875.
[73] S. Helal, R. Najee, M. Hanna, M. Shaaban, A. Osman, M. Hassan, An
energy management system for hybrid microgrids in remote commu-
nities, in: Electrical and Computer Engineering (CCECE), 2017 IEEE
30th Canadian Conference on, IEEE, 2017, pp. 1–4.
[74] A. G. Tsikalakis, N. D. Hatziargyriou, Centralized control for optimiz-
ing microgrids operation, in: Power and Energy Society General Meet-
ing, 2011 IEEE, IEEE, 2011, pp. 1–8.
[75] B. Heymann, J. F. Bonnans, P. Martinon, F. J. Silva, F. Lanas,
G. Jim´
enez-Est´
evez, Continuous optimal control approaches to micro-
grid energy management, Energy Systems (2015) 1–19.
[76] L. N. An, T. Quoc-Tuan, Optimal energy management for grid connected
microgrid by using dynamic programming method, in: Power & Energy
Society General Meeting, 2015 IEEE, IEEE, 2015, pp. 1–5.
[77] M. Strelec, J. Berka, Microgrid energy management based on approx-
imate dynamic programming, in: Innovative Smart Grid Technologies
Europe (ISGT EUROPE), 2013 4th IEEE/PES, IEEE, 2013, pp. 1–5.
[78] H. Kanchev, D. Lu, F. Colas, V. Lazarov, B. Francois, Energy manage-
ment and operational planning of a microgrid with a pv-based active
generator for smart grid applications, IEEE transactions on industrial
electronics 58 (10) (2011) 4583–4592.
[79] A. Merabet, K. T. Ahmed, H. Ibrahim, R. Beguenane, A. M. Ghias,
Energy management and control system for laboratory scale microgrid
based wind-pv-battery, IEEE Transactions on Sustainable Energy 8 (1)
(2017) 145–154.
[80] A. Choudar, D. Boukhetala, S. Barkat, J.-M. Brucker, A local energy
management of a hybrid pv-storage based distributed generation for mi-
crogrids, Energy Conversion and Management 90 (2015) 21–33.
[81] J. Pascual, J. Barricarte, P. Sanchis, L. Marroyo, Energy management
strategy for a renewable-based residential microgrid with generation and
demand forecasting, Applied Energy 158 (2015) 12–25.
[82] M. Sechilariu, B. Wang, F. Locment, Building-integrated microgrid:
Advanced local energy management for forthcoming smart power grid
communication, Energy and Buildings 59 (2013) 236–243.
[83] M. Elsied, A. Oukaour, T. Youssef, H. Gualous, O. Mohammed, An
advanced real time energy management system for microgrids, Energy
114 (2016) 742–752.
[84] S. Chalise, J. Sternhagen, T. M. Hansen, R. Tonkoski, Energy manage-
ment of remote microgrids considering battery lifetime, The Electricity
Journal 29 (6) (2016) 1–10.
[85] S. Golshannavaz, S. Afsharnia, P. Siano, A comprehensive stochastic
energy management system in reconfigurable microgrids, International
Journal of Energy Research 40 (11) (2016) 1518–1531.
[86] J. Radosavljevi´
c, M. Jevti´
c, D. Klimenta, Energy and operation man-
agement of a microgrid using particle swarm optimization, Engineering
Optimization 48 (5) (2016) 811–830.
[87] H. Li, A. T. Eseye, J. Zhang, D. Zheng, Optimal energy management for
industrial microgrids with high-penetration renewables, Protection and
Control of Modern Power Systems 2 (1) (2017) 12.
[88] S. A. Alavi, A. Ahmadian, M. Aliakbar-Golkar, Optimal probabilistic
energy management in a typical micro-grid based-on robust optimiza-
tion and point estimate method, Energy Conversion and Management
95 (2015) 314–325.
[89] M. Manbachi, M. Ordonez, Ami-based energy management for islanded
ac/dc microgrids utilizing energy conservation and optimization, IEEE
Transactions on Smart Grid PP (99) (2017) 1.
[90] V. Mohan, J. G. Singh, W. Ongsakul, An efficient two stage stochastic
optimal energy and reserve management in a microgrid, Applied Energy
160 (2015) 28–38.
[91] N. Tiwari, L. Srivastava, Generation scheduling and micro-grid energy
management using differential evolution algorithm, in: Circuit, Power
and Computing Technologies (ICCPCT), 2016 International Conference
on, IEEE, 2016, pp. 1–7.
[92] M. Yu, Y. Wang, Y. Li, Energy management of wind turbine-based dc
microgrid utilizing modified differential evolution algorithm, in: Inter-
national Conference on Renewable Power Generation (RPG 2015), IET,
2015, pp. 1–6.
[93] M. Marzband, E. Yousefnejad, A. Sumper, J. L. Dom´
ınguez-Garc´
ıa,
Real time experimental implementation of optimum energy management
system in standalone microgrid by using multi-layer ant colony opti-
mization, International Journal of Electrical Power & Energy Systems
75 (2016) 265–274.
[94] M. Marzband, M. Ghadimi, A. Sumper, J. L. Dom´
ınguez-Garc´
ıa, Ex-
perimental validation of a real-time energy management system using
multi-period gravitational search algorithm for microgrids in islanded
mode, Applied energy 128 (2014) 164–174.
[95] T. Niknam, F. Golestaneh, A. Malekpour, Probabilistic energy and oper-
ation management of a microgrid containing wind/photovoltaic/fuel cell
generation and energy storage devices based on point estimate method
and self-adaptive gravitational search algorithm, Energy 43 (1) (2012)
427–437.
[96] M. Motevasel, A. R. Seifi, Expert energy management of a micro-grid
considering wind energy uncertainty, Energy Conversion and Manage-
ment 83 (2014) 58–72.
[97] M. Marzband, S. S. Ghazimirsaeid, H. Uppal, T. Fernando, A real-time
evaluation of energy management systems for smart hybrid home micro-
grids, Electric Power Systems Research 143 (2017) 624–633.
[98] M. Marzband, F. Azarinejadian, M. Savaghebi, J. M. Guerrero, An opti-
mal energy management system for islanded microgrids based on multi-
period artificial bee colony combined with markov chain, IEEE Systems
Journal 11 (3) (2017) 1712–1722.
[99] S. Mohammadi, S. Soleymani, B. Mozafari, Scenario-based stochas-
tic operation management of microgrid including wind, photovoltaic,
micro-turbine, fuel cell and energy storage devices, International Jour-
nal of Electrical Power & Energy Systems 54 (2014) 525–535.
[100] R. Velik, P. Nicolay, Grid-price-dependent energy management in mi-
crogrids using a modified simulated annealing triple-optimizer, Applied
Energy 130 (2014) 384–395.
[101] B. Papari, C. Edrington, T. Vu, F. Diaz-Franco, A heuristic method
for optimal energy management of dc microgrid, in: DC Microgrids
(ICDCM), 2017 IEEE Second International Conference on, IEEE, 2017,
pp. 337–343.
[102] K. Huang, X. Wang, L. Wang, Optimal energy management of grid-
connected photovoltaic micro-grid, in: Cyber Technology in Automa-
tion, Control, and Intelligent Systems (CYBER), 2015 IEEE Interna-
tional Conference on, IEEE, 2015, pp. 234–239.
[103] M. Marzband, N. Parhizi, J. Adabi, Optimal energy management for
stand-alone microgrids based on multi-period imperialist competition
algorithm considering uncertainties: experimental validation, Interna-
tional transactions on electrical energy systems 26 (6) (2016) 1358–
1372.
[104] D. Arcos-Aviles, J. Pascual, F. Guinjoan, L. Marroyo, P. Sanchis, M. P.
Marietta, Low complexity energy management strategy for grid profile
smoothing of a residential grid-connected microgrid using generation
and demand forecasting, Applied Energy 205 (2017) 69–84.
[105] Y.-K. Chen, Y.-C. Wu, C.-C. Song, Y.-S. Chen, Design and implementa-
tion of energy management system with fuzzy control for dc microgrid
systems, IEEE Transactions on Power Electronics 28 (4) (2013) 1563–
1570.
[106] G. Kyriakarakos, A. I. Dounis, K. G. Arvanitis, G. Papadakis, A fuzzy
logic energy management system for polygeneration microgrids, Re-
newable Energy 41 (2012) 315–327.
[107] E. De Santis, A. Rizzi, A. Sadeghian, Hierarchical genetic optimization
of a fuzzy logic system for energy flows management in microgrids,
Applied Soft Computing 60 (2017) 135–149.
[108] T. Wang, X. He, T. Deng, Neural networks for power management op-
timal strategy in hybrid microgrid, Neural Computing and Applications
(2017) 1–13.
[109] M. E. G. Urias, E. N. Sanchez, L. J. Ricalde, Electrical microgrid op-
23
timization via a new recurrent neural network, IEEE Systems Journal
9 (3) (2015) 945–953.
[110] G. K. Venayagamoorthy, R. K. Sharma, P. K. Gautam, A. Ahmadi, Dy-
namic energy management system for a smart microgrid, IEEE transac-
tions on neural networks and learning systems 27 (8) (2016) 1643–1656.
[111] E. Kuznetsova, Y.-F. Li, C. Ruiz, E. Zio, G. Ault, K. Bell, Reinforcement
learning for microgrid energy management, Energy 59 (2013) 133–146.
[112] S. Ghorbani, R. Rahmani, R. Unland, Multi-agent autonomous decision
making in smart micro-grids energy management: A decentralized ap-
proach, in: German Conference on Multiagent System Technologies,
Springer, 2017, pp. 223–237.
[113] A. Anvari-Moghaddam, A. Rahimi-Kian, M. S. Mirian, J. M. Guerrero,
A multi-agent based energy management solution for integrated build-
ings and microgrid system, Applied Energy 203 (2017) 41–56.
[114] H. K. Nunna, S. Doolla, Energy management in microgrids using de-
mand response and distributed storagea multiagent approach, IEEE
Transactions on Power Delivery 28 (2) (2013) 939–947.
[115] C.-X. Dou, B. Liu, Multi-agent based hierarchical hybrid control for
smart microgrid, IEEE transactions on smart grid 4 (2) (2013) 771–778.
[116] C.-S. Karavas, G. Kyriakarakos, K. G. Arvanitis, G. Papadakis, A multi-
agent decentralized energy management system based on distributed in-
telligence for the design and control of autonomous polygeneration mi-
crogrids, Energy Conversion and Management 103 (2015) 166–179.
[117] N. Liu, X. Yu, C. Wang, J. Wang, Energy sharing management for mi-
crogrids with pv prosumers: A stackelberg game approach, IEEE Trans-
actions on Industrial Informatics 13 (3) (2017) 1088–1098.
[118] L. Ma, N. Liu, J. Zhang, W. Tushar, C. Yuen, Energy management for
joint operation of chp and pv prosumers inside a grid-connected mi-
crogrid: A game theoretic approach, IEEE Transactions on Industrial
Informatics 12 (5) (2016) 1930–1942.
[119] G. E. Asimakopoulou, A. L. Dimeas, N. D. Hatziargyriou, Leader-
follower strategies for energy management of multi-microgrids, IEEE
Transactions on Smart Grid 4 (4) (2013) 1909–1916.
[120] F. A. Mohamed, H. N. Koivo, Multiobjective optimization using mod-
ified game theory for online management of microgrid, International
Transactions on Electrical Energy Systems 21 (1) (2011) 839–854.
[121] Y. Lan, X. Guan, J. Wu, Rollout strategies for real-time multi-energy
scheduling in microgrid with storage system, IET Generation, Trans-
mission & Distribution 10 (3) (2016) 688–696.
[122] K. Jia, Y. Chen, T. Bi, Y. Lin, D. Thomas, M. Sumner, Historical-data-
based energy management in a microgrid with a hybrid energy stor-
age system, IEEE Transactions on Industrial Informatics 13 (5) (2017)
2597–2605.
[123] A. Ghasemi, Coordination of pumped-storage unit and irrigation system
with intermittent wind generation for intelligent energy management of
an agricultural microgrid, Energy 142 (2018) 1–13.
[124] G. Cau, D. Cocco, M. Petrollese, S. K. Kær, C. Milan, Energy manage-
ment strategy based on short-term generation scheduling for a renew-
able microgrid using a hydrogen storage system, Energy Conversion and
Management 87 (2014) 820–831.
[125] W. Su, J. Wang, J. Roh, Stochastic energy scheduling in microgrids with
intermittent renewable energy resources, IEEE Transactions on Smart
Grid 5 (4) (2014) 1876–1883.
[126] J. Shen, C. Jiang, Y. Liu, X. Wang, A microgrid energy management
system and risk management under an electricity market environment,
IEEE Access 4 (2016) 2349–2356.
[127] M.-C. Hu, S.-Y. Lu, Y.-H. Chen, Stochastic programming and market
equilibrium analysis of microgrids energy management systems, Energy
113 (2016) 662–670.
[128] H. Farzin, M. Fotuhi-Firuzabad, M. Moeini-Aghtaie, Stochastic energy
management of microgrids during unscheduled islanding period, IEEE
Transactions on Industrial Informatics 13 (3) (2017) 1079–1087.
[129] J. Liu, H. Chen, W. Zhang, B. Yurkovich, G. Rizzoni, Energy man-
agement problems under uncertainties for grid-connected microgrids:
a chance constrained programming approach, IEEE Transactions on
Smart Grid 8 (6) (2017) 2585–2596.
[130] E. Kuznetsova, Y.-F. Li, C. Ruiz, E. Zio, An integrated framework of
agent-based modelling and robust optimization for microgrid energy
management, Applied Energy 129 (2014) 70 – 88.
[131] Y. Xiang, J. Liu, Y. Liu, Robust energy management of microgrid with
uncertain renewable generation and load, IEEE Transactions on Smart
Grid 7 (2) (2016) 1034–1043.
[132] W. Hu, P. Wang, H. Gooi, Towards optimal energy management of mi-
crogrids via robust two-stage optimization, IEEE Transactions on Smart
Grid 9 (2) (2018) 1161–1174.
[133] Y. Zhang, N. Gatsis, G. B. Giannakis, Robust energy management
for microgrids with high-penetration renewables, IEEE Transactions on
Sustainable Energy 4 (4) (2013) 944–953.
[134] L. Zhao, G. Wei, W. Zhi, W. Zhihe, T. Yiyuan, A robust optimization
method for energy management of cchp microgrid, Journal of Modern
Power Systems and Clean Energy (2017) 1–13.
[135] B. V. Solanki, A. Raghurajan, K. Bhattacharya, C. A. Canizares, Includ-
ing smart loads for optimal demand response in integrated energy man-
agement systems for isolated microgrids, IEEE Transactions on Smart
Grid 8 (4) (2017) 1739–1748.
[136] P. R. Mendes, L. V. Isorna, C. Bordons, J. E. Normey-Rico, Energy man-
agement of an experimental microgrid coupled to a v2g system, Journal
of Power Sources 327 (2016) 702–713.
[137] L. I. Minchala-Avila, L. Garza-Casta˜
n´
on, Y. Zhang, H. J. A. Ferrer, Op-
timal energy management for stable operation of an islanded microgrid,
IEEE Transactions on Industrial Informatics 12 (4) (2016) 1361–1370.
[138] I. Prodan, E. Zio, A model predictive control framework for reliable
microgrid energy management, International Journal of Electrical Power
& Energy Systems 61 (2014) 399–409.
[139] F. Valencia, J. Collado, D. S´
aez, L. G. Mar´
ın, Robust energy manage-
ment system for a microgrid based on a fuzzy prediction interval model,
IEEE Transactions on Smart Grid 7 (3) (2016) 1486–1494.
[140] Z. Luo, Z. Wu, Z. Li, H. Cai, B. Li, W. Gu, A two-stage optimization
and control for cchp microgrid energy management, Applied Thermal
Engineering 125 (2017) 513–522.
[141] B. V. Solanki, K. Bhattacharya, C. A. Ca˜
nizares, A sustainable energy
management system for isolated microgrids, IEEE Transactions on Sus-
tainable Energy 8 (4) (2017) 1507–1517.
[142] I. Prodan, E. Zio, F. Stoican, Fault tolerant predictive control design for
reliable microgrid energy management under uncertainties, Energy 91
(2015) 20–34.
[143] L. Guo, W. Liu, X. Li, Y. Liu, B. Jiao, W. Wang, C. Wang, F. Li, Energy
management system for stand-alone wind-powered-desalination micro-
grid, IEEE Transactions on Smart Grid 7 (2) (2016) 1079–1087.
[144] X. Jin, J. Wu, Y. Mu, M. Wang, X. Xu, H. Jia, Hierarchical microgrid
energy management in an office building, Applied Energy 208 (2017)
480–494.
[145] M. Akter, M. Mahmud, A. M. Oo, A hierarchical transactive energy
management system for microgrids, in: Power and Energy Society Gen-
eral Meeting (PESGM), 2016, IEEE, 2016, pp. 1–5.
[146] F. F. Yanine, F. M. Crdova, L. Valenzuela, Sustainable hybrid energy
systems: An energy and exergy management approach with homeostatic
control of microgrids, Procedia Computer Science 55 (2015) 642 – 649,
3rd International Conference on Information Technology and Quantita-
tive Management, ITQM 2015.
[147] F. F. Yanine, F. I. Caballero, E. E. Sauma, F. M. Crdova, Building sus-
tainable energy systems: Homeostatic control of grid-connected micro-
grids, as a means to reconcile power supply and energy demand response
management, Renewable and Sustainable Energy Reviews 40 (2014)
1168 – 1191.
[148] W. Shi, X. Xie, C.-C. Chu, R. Gadh, Distributed optimal energy man-
agement in microgrids, IEEE Transactions on Smart Grid 6 (3) (2015)
1137–1146.
[149] F. A. Mohamed, H. N. Koivo, System modelling and online optimal
management of microgrid using mesh adaptive direct search, Interna-
tional Journal of Electrical Power & Energy Systems 32 (5) (2010) 398–
407.
[150] https://www.schneider-electric.com/en/work/products/
medium-voltage- switchgear-and- energy-automation/
r-and- d-projects/migrate.jsp, accessed: 12-04-2018.
[151] http://new.abb.com/distributed-energy- microgrids/
our-offering/microgrid- plus-system, accessed: 12-04-2018.
[152] https://www.gegridsolutions.com/multilin/catalog/mcs.
htm, accessed: 12-04-2018.
[153] https://www.siemens.com/global/en/home/products/
energy/energy-automation- and-smart- grid/microgrid/
spectrum-power- mgms.html, accessed: 13-04-2018.
24
