IEEE POWER & ENERGY SOCIETY SECTION

Received 21 June 2022, accepted 14 July 2022, date of publication 20 July 2022, date of current version 27 July 2022.
Digital Object Identifier 10.1109/ACCESS.2022.3192638

A Review on Optimal Energy Management of

Multimicrogrid System Considering Uncertainties
GUOLONG MA1 , JIANING LI 2, (Member, IEEE), AND XIAO-PING ZHANG 1, (Fellow, IEEE)
1 Departmentof Electronic, Electrical and Systems Engineering, University of Birmingham, Birmingham B15 2TT, U.K.
2 WSP U.K., The Mailbox, Birmingham B1 1RQ, U.K.

Corresponding author: Jianing Li ([email protected])

This work was supported in part by the Engineering and Physical Sciences Research Council (EPSRC) under Grant EP/N032888/1 and
Grant EP/L017725/1, and in part by the China Scholarship Council.

ABSTRACT Microgrid (MG) is one of the most effective solution to integrate distributed renewable energy
into power system. However, modern MG has several technical challenges such as migration to multi vector
energy system, increasing renewable energy penetration, and the uncertainties that arise from the large-
scale incorporation of renewable energy. Recent development of multi-microgrid (MMG) could potentially
mitigate these challenges. MMG has obvious advantages in improving the system stability, reliability,
economy, and energy efficiency of power grid operation through autonomous management and coordinated
control among networked MGs. In the meantime, uncertainties from many elements and operators in the
MMG system also pose huge challenges in modeling. This paper will present and review typical architecture
of MMG, including physical layer, information layer and application layer. Moreover, this paper will
critically review and analyze challenges in uncertainty modelling and solution in MMGs. Finally, the paper
will also discuss future research areas and development trends of MMG.

INDEX TERMS Microgrid, multi-microgrid system, energy management, optimization, architecture,

uncertainties.

I. INTRODUCTION IMARC, the global MG market size will reach $25.9 billion
Due to the rising public awareness in environmental con- in 2021and the market will reach $49.2 billion by 2027, with
cerns, several countries and organizations have established a CAGR of 11.2% from 2022-2027. As shown in Figure 1.
legal commitments in delivering a more suitable energy sys-
tem [1], [2]. Meanwhile, the volatility in electricity price
caused by reduction of fossil fuel reserves, renewable energy
generation has become the most popular solution to meet
energy demand in future energy scenarios [3]. Particularly,
in the electricity sector, wind and photovoltaic power are
experiencing the highest growth in different countries. Owing
to their intermittent and distributed nature, the integration
of large-scale renewable energy in conventional way is a
challenging process. MG is one of the most effective solution
to integrate distributed renewable energy into power system,
which usually consist of confined cluster of loads, energy
FIGURE 1. Global microgrid market size (2021-2027).
storages and small generators [4]. Based on the reports from
However, MG has several technical challenges, including
The associate editor coordinating the review of this manuscript and migration to multi vector energy system, increasing renew-
approving it for publication was Yonggang Liu . able energy penetration, and the uncertainties that arise from

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
VOLUME 10, 2022 77081
 G. Ma et al.: Review on Optimal Energy Management of MMG System Considering Uncertainties

the large-scale incorporation of renewable energy. Connect- varying photovoltaic and wind power generation, charging/
ing multiple MGs to construct multi-microgrid system could discharging behavior of electric vehicles, stochastic topol-
mitigate these challenges. MMG refers to a grid formed by ogy of MMG systems and forecasted electricity price, etc.
interconnecting two or more single MGs through a com- [28]–[30]. Based on these uncertainties, the security, reliabil-
mon coupling point, and this interconnection enables power ity, and economy of MMG system will be deeply affected.
interaction between multiple MGs or with the upper-level To have a realistic modeling and making better decisions in
grid [5]. According to the IEEE Standard 1547.4 [6], the MMG systems, the uncertainties must be considered. Con-
operation and reliability of a distribution system can be sidering the influence of the above uncertain factors, the
improved by splitting it into multiple MGs. MMG’s technical optimal energy management of MMG system becomes the
advantages are mainly reflected in the following aspects: optimization decision problem with multiple uncertainties.
1) in MMG, the energy sharing among the MGs allows The common approaches for dealing with these uncertainty
the MGs to satisfy their power demand with their own problems can be classified into three main categories. In the
cheaper renewable energy resource which can reduce the cost approaches of the first category, referred to as stochastic
of fossil fueled generation [7]; 2) the interconnected MGs programming, which need to assume uncertainty variables
will effectively improve the utilization of renewable energy, follow a predefined probability distribution and then we can
reduce the burden on the main grid, and can also improve transform it into a deterministic mathematical programming
the reliability not only for themselves but also for the main problem and solve it [31]. In the approaches of the second cat-
grid [8]; 3) in MMG, each microgrid has high self-healing egory, referred to as robust optimization, which can express
feature and it can also operate in a stable way by sharing the fluctuation range of the uncertain parameters by setting
extra energy and spare resources with other connected MGs up an uncertain set and the robust optimization can gener-
[9]–[12]; 4) in MMG, the interconnected MGs can support ally obtain feasible solutions [32], [33]. The third category
each other with local generation capacities to achieve the called distributionally robust optimization which remedies
overall reliability and minimize the total load curtailment in the stochastic optimization’s concreteness of probability
the system [13], [14]. Based on these technical advantages, information and robust optimization’s ignorance of probabil-
MMG has been used in practical applications, reference [15] ity information is proposed to solve the uncertainties in MMG
introduced the Illinois Institute of Technology DC MMG, system [34], [35].
which mainly is used for economic dispatching analysis. To provide a look-up reference for fledgling researchers
In [16], U.S. Department of Energy want to design a master on the status of MMG research, this paper summarizes the
controller for achieving a seamless integration of Bronzeville review papers on MMG related research in recent years,
community microgrid and the Illinois Institute of Tech- which mainly focusing on the following issues: 1) MMG
nology campus microgrid. What’s more, a company called architectures [14], [24], [36]–[38]; 2) MMG communica-
CESI built MMG simulation platform consisting of super tion [24]; 3) energy management of MMG [37], [38]; 4) oper-
capacitor storage and flywheel energy storage, is beneficial ation, protection, and resilience of MMG [12], [24], [36],
to energy quality analysis, communication, and the upper etc. Reference [12], [24], [38] examine the possible multi-
control in MG [14]. microgrid architectures to form a grid of microgrids, and there
Meanwhile, for the stable and economic operation of are some comparisons between the different architectures
MMG, a multi-microgrid energy management system is is performed in terms of operation, protection, reliability,
required to manage and coordinate dispatchable distributed stability, communications, and energy management in [24],
generators, controllable loads, and energy trading among [36]–[38]. But the above review papers deal with the uncer-
networked MGs. The operation objectives of MMG can be tainty problems in the MMG relatively roughly and lack a
summarized as follows: 1) to minimize the overall operation comprehensive analysis of uncertainty impact on the oper-
cost of MMG in the grid-connected mode [17] and islanded ation control and energy management of the MMG. Based
mode [18]; 2) to maximize the distributed renewable energy on these challenges, the thesis of this paper is to examine
penetration with the influence of uncertainties from various the literature on the uncertainties of energy management in
DREs, loads and electricity price [19]–[21]; 3) to minimize MMG system in recognition of the detrimental impacts of dif-
energy transmission loss from energy exchanges with the ferent uncertainties including operation, energy market and
main-grid and among networked MGs [22]; 4) to maximize power grid faults on energy management. On the one hand,
the economic goals of different subjects by designing multi- this paper aims to provide a look-up reference for fledgling
microgrids cluster architecture and its energy transaction researchers on the status of MMG research and can also help
mechanism [23]; 5) to minimize the communication delay researchers to establish a theoretical framework quickly and
between MMGs and improve the security and reliability of accurately for energy management of MMG by citing related
communication system [24]. articles on the operation and management of MMG; On the
Although MMG has obvious advantages, due to its rel- other hand, this paper aims to analyze the characteristics and
atively complex structure and large number of elements, influences of different uncertainties and then to provide a
it still faces great challenges suffering from uncertain- complete solution about the operation, modeling, and control
ties [25]–[27]. These uncertainties include load forecasting, for the energy management of MMG.

77082 VOLUME 10, 2022

G. Ma et al.: Review on Optimal Energy Management of MMG System Considering Uncertainties

FIGURE 2. Energy management architecture of MMG.

In summary, the contributions of this paper are as follows. Based on the above idea, this paper proposes that the
• The typical architecture and functionalities of MMG, energy management framework of MMG can be divided into
including physical layer, information layer and appli- three levels including physical layer, information layer and
cation layer, which includes energy management strat- application layer, and the framework is shown in Figure 2.
egy, energy operation control method, and energy The first one is the physical layer, which mainly includes
management method with uncertainties, are thoroughly the power generation units, electrical equipment, energy stor-
analyzed. age equipment and other physical equipment of the MG
• The main challenging problem of MMG is suffering [41], [42]. The second layer is the information layer of
from all kinds of uncertainty elements. This paper ana- MMG, which mainly includes state of the system equip-
lyzes the uncertainties and methods to deal with these ment, energy market information and network information,
uncertainties in power system analysis, operation, and etc. This layer mainly realizes the collection, storage, and
control. process of information and data. The third layer is the appli-
• Various model and solving method in this paper are cation layer, which contains the energy management strategy,
clearly introduced to deal with MMG with various energy operation control method and energy management
uncertainties, including statistic optimization, robust method with uncertainties. Energy management strategy is
optimization and distributionally robust optimization are mainly to select different control strategies of MMG [43].
thoroughly analyzed. And energy management control method has main function
The remainder of this research paper is organized as fol- of power coordination and economic dispatching [44], [45].
lows: Section II provides a comprehensive overview on the
typical architecture of MMG. The uncertain challenges in
A. ENERGY MANAGEMENT STRATEGY
MMG are surveyed in Section III. Section IV investigates the
The main difference between MMG and single MG is that
model and solving method of energy scheduling for MMG
MMG needs to further consider the coordination strategy
with uncertainties. Section V discusses several challenging
in different MGs, because under different control strate-
problems for the future development of MMG. Finally, the
gies, MMG has different communication network require-
conclusion is presented in Section VI.
ment [46], [47]. Reasonable strategy design plays a key role
II. ENERGY MANAGEMENT ARCHITECTURE OF
in simplifying the control complexity of MMG system and
MULTI-MICROGRID SYSTEM
improving the economy of system operation. At present,
The energy management of MMG system is the core of there are mainly three control strategies for the research of
ensuring the economic and stable operation of the system. MMG [48].
The main function of the architecture is to collect, process
and analyze various kinds of information, such as the state of 1) CENTRALIZED
the system equipment, market information, etc. and on this Centralized strategy is the centralized management of each
basis, make decisions and control on the real-time operating single MG through a central energy management sys-
state of MMG [39], [40]. tem [8], [49]. In this strategy, each MG connects to the main

VOLUME 10, 2022 77083

 G. Ma et al.: Review on Optimal Energy Management of MMG System Considering Uncertainties

respectively [53], [54]. In particular, 1) when one of the MGs

failures or energy supply and demand is unbalanced, this MG
can communicate with the neighboring MGs and main grid
through the internal energy management system, and then
the networked MGs/main grid can supply energy support,
so that it can operate stably again [24]; 2) when the main grid
fails and all the MG are disconnected from main grid, that is,
when multiple MGs are in an islanded mode, the MGs can
also realize energy interaction through the interconnection
between them [55]. A significant difference regarding the
FIGURE 3. Centralized control strategy of MMG. other strategy is that individual agents are allowed to discover
or exchange other agent information through communication
and coordination with their neighbors.
grid directly. As shown in Figure 3, there is no energy trading As shown in Figure 4, each MG is fully connected with
directly among the MGs. It shows the local energy system neighboring MGs, its stability will be the highest, but this
consisting of n MGs, as well as the information exchange strategy has higher control complexity and poor economy.
among the MGs and a central energy management system Although stability of distributed control strategy is supe-
(CEMS). Each MG is allowed to respond to the decision rior to the centralized one, but distributed also exists some
of the CEMS and its own energy management system. The shortcomings, due to the lack of global information, this
energy sharing and trading mainly rely on dedicated power control strategy is difficult to achieve the global optimization
exchange lines. For instance, if one of the MGs is unable goal [56]. Specifically, when the system wants to get the
to satisfy its own load demand, it can purchase electrical global optimization result, it must iterate many times until
power from the main grid through the bus line. On the other convergence. Even though, the global optimal solution may
hand, an MG with surplus energy can benefit by selling not be obtained.
the extra power to the main grid through the bus line. The
main advantage of centralized strategy is its simple structure,
which can use the central energy manager to deal with the
interaction information of each single MG in a centralized
manner, without setting complex communication rules and
energy interaction protocols [50]. However, as the number of
MG increases, there will be a sharp increase in information
exchange, data volume and energy interaction frequency, and
the computing speed of the CEMS may not meet the real-time
requirements of power system scheduling [51]. In addition,
if a single MG has a connection failure, it cannot be sup-
ported by the other MGs. Moreover, centralized strategy still
has scalability limits, performance degradation and lack of FIGURE 5. Hybrid control strategy of MMG.
robustness [52].

3) HYBRID
The MMG adopt hybrid strategy can realize energy and infor-
mation interaction and sharing by interconnected with adja-
cent MGs [57]. Hybrid strategy combines the advantages of
centralized strategy and distributed strategy. Mainly reflected
in the following aspects: 1) Can easily achieve the global
optimization. In hybrid strategy, the CEMS controller can get
the global information, which can help reduce the operation
cost [58]. 2) Reduction on communication and computation
burden of CEMS controller. In hybrid strategy, it can effec-
FIGURE 4. Distributed control strategy of MMG. tively alleviate the performance bottlenecks caused by the
central energy router in the centralized strategy, such as infor-
mation congestion caused by excessive information, user’s
2) DISTRIBUTED privacy disclosure, poor real-time performance, etc. [24].
In the distributed strategy, each MG is locally intercon- The hybrid strategy structure, as shown in figure 5, can
nected through a certain connection mode and realizes effectively reduce the communication and calculation time
energy and information interaction with the main grid of the central energy manager, which is helpful for real-time

77084 VOLUME 10, 2022

G. Ma et al.: Review on Optimal Energy Management of MMG System Considering Uncertainties

TABLE 1. ROS and cons for the three management strategies of MMG.

energy scheduling [59], [60]. When one of the MG failure or

energy imbalance occur, the requirement information could
be sent to the neighboring MG, and then obtain energy sup-
port from the adjacent MGs; If the adjacent MG cannot meet
the demand, the sub-energy management system can transfer
information to the CEMS and then the CEMS can coordinate
to complete energy exchange through the sub-microgrid sys-
tem in other areas. FIGURE 6. Power energy sharing mechanism.
Table 1 is mainly to compared advantages and disadvan-
tages of different energy management strategies. Specially,
the MMG will be connected to many distributed renewable amount of electricity generated (shortage) by MG after its
energy and power loads in the future, the selection of MMG local optimization. Specially, we assume that the electricity
control strategy will mainly consider its economy, stability, price in different MG is different.
scalability, communication complexity, control complexity From the figure 6, in the centralized mechanism each MG
and other indicators [38]. Economy is largely determined can only exchange the energy with the main grid [72]–[74].
by infrastructure and operating cost. Infrastructure cost will In the distributed mechanism, MG1, MG2 and MG3 can form
depend on the number of components, their power rating and a cluster, which can realize the whole cluster optimization.
the technology used, while operating cost will be affected In this cluster, MG1 and MG2 can send the surplus amount
by factors such as generation cost, ancillary service cost or to fulfill shortage amount of MG3. The remaining surplus of
power transmission loss; Stability refers to the ability of the this cluster can be sent to the main grid and the MGs in this
MMG to recover to a stable state after being disturbed, that cluster can obtain profit by selling the surplus energy. In the
is, whether a MG can quickly get the energy support from hybrid mechanism, each MG can make energy transaction
other MGs to restore stable operation when energy shortage with the other MGs and the main grid [56]. From Figure 6(c),
or system failure occur. Scalability refers to the ability of we can know that MG2 can send the surplus to the main grid,
the system to accept new elements, such as the ability to and it can also to fulfill shortage of MG3, which one can get
accept the internal distributed system of the existing MG, the the surplus energy depending on the money who can give
ability of load growth and the ability to accept the access more; at the same time, MG1 can also fulfill shortage of MG3
of the new MG. Communication complexity refers to the (generation cost of MG1 is lowest). Especially, when the
size of communication volume and communication delay remaining surplus amount of MG2 is sent to MG1, and then
time, etc. Data privacy represents the data or information MG1 can reduce its generation (generation cost of MG1 >
leakage in energy interaction or transaction in MMG. Control selling price). Similarly, when shortage of MG3 in Figure 6(c)
complexity refers to the difficulty of setting control policies cannot be fulfilled only by MG2. MG3 must buy the remain-
for different topologies to achieve stable operation. ing shortage from MG1 (MG1’s generation cost < buying
price from the main grid). MG1 can increase its generation
and send to MG3. In short, under this transaction mechanism,
B. POWER ENERGY SHARING MECHANISMS AND we can maximize the benefits of the whole society, and we
CONSTRAINTS can also make the best use of the power generation resources
The interaction mechanism between information and energy from each MG.
depends on the network topology, which including MMG
physical topology and communication network [71]. In this C. ENERGY MANAGEMENT STRATEGY AND OPERATION
section, information, and energy interaction mechanisms in CONTROL METHOD FOR MULTI-MICROGRID SYSTEM
MMG are shown in Figure 6. The energy management strategy of MMG is mainly to
The dotted lines represent the power generated in MGs coordinate different MGs and superior distribution network,
is equal to the load demands. The light blue color depicts to maximize the use of renewable energy, improve the sta-
the amount of electricity generated (surplus) by MG after bility and economy of the whole system. The cooperative
its local optimization. The light orange color depicts the approaches in the existing studies are mainly clarified into

VOLUME 10, 2022 77085

 G. Ma et al.: Review on Optimal Energy Management of MMG System Considering Uncertainties

TABLE 2. Energy management scheduling strategy and operation control method for MMG.

three categories, namely centralized schemes, distributed compared the different solution methods and optimization
approaches, and hybrid control strategies [57]. The central- objectives of energy management and operation control of
ized schemes work well for conventional power system by MMG under different cooperative approaches.
introducing a MG central controller. The methods imple-
mented in a centralized manner require an information center 1) ENERGY TRANSACTION STRATEGY
to collect global information and a central controller to pro- Energy transaction strategy for MMG often involves multi-
cess the amounts of received data. Thus, the required com- partner, which can be solved by game theory, as an advanced
putational capacity of the control center is rapidly growing optimization tool, is mainly used to study how multiple stake-
with the increase of the power devices, and it may be more holders make optimization decisions and has been widely
vulnerable to the single-point failures [52]. The distributed used in the field of power system. Energy interaction and
approaches may be robust and less costs in terms of no cooperative control methods between MMG can generally be
communication network are needed. However, in [8], it shows transformed into cooperative game model or non-cooperative
that the available resources in the network would be not game model [75]–[77]. Reference [75] discussed the possi-
utilized in a cost-effective way because of the deficiency bility of MMG cooperating through coordinated economic
of broader available information. In contrast, the coopera- scheduling, and proposed Nash-Harsanyi cost allocation
tive solution in the distributed way that only utilizes local scheme under the framework of cooperative game theory
information through a local private communication network. to ensure fair and stable cost sharing among MGs. Refer-
In the future development, it will consist of more hierar- ence [76] proposed a scheme for the collaborative energy
chy distributed controllable power-electronics devices with and reserve scheduling model to participate in the optimal
the ability to exchange information through a communica- operation of MMG. However, in most cases, it is difficult
tion network. Therefore, the emerging management solution for MMG cooperative operation to simultaneously meet the
should be efficient and low-cost for an economically viable profits of each MG, so it is necessary to set power transaction
smart grid. The main research directions of current MMG strategies based on non-cooperative game according to the
energy management and operation control in different coop- preferences of decision makers. Reference [77] developed
erative approaches can be summarized as the following three a non-corporative games theory model based on distributed
aspects:1) energy transaction strategy [75]–[79]; 2) informa- energy management algorithm for multiple smart distribution
tion interaction strategy [80]–[83]; 3) artificial intelligence systems of MG including real-time electricity price, which
technology in MMG [87]–[93]. Table 2 summarized and can effectively coordinate the operation of MGs and achieve

77086 VOLUME 10, 2022

G. Ma et al.: Review on Optimal Energy Management of MMG System Considering Uncertainties

energy balance among different MGs. Reference [78] pro- Then the model-free reinforcement learning technique is
posed a decentralized control model to improve the fault used to optimize the retail price of local MG to maximize
ride-through of a multi-microgrid system. Reference [79] the profit by selling energy. Reference [90] proposes an
introduce an energy interaction framework for a decentralized Internet of Things (IoT) platform for energy management
operation in a stand-alone multi microgrid system to improve in multi-microgrid system to enhance the power quality.
the economic and resilience. Reference [91] design a framework of integrated frequency
control to reduce the frequency deviation of a multi-area
2) INFORMATION INTERACTION STRATEGY multi-microgrid system. In addition, traditional artificial
Aiming at the information interaction strategy of MMG, intelligence algorithms, such as particle swarm optimization
researchers put forward an information interaction strategy and genetic algorithm, have obvious advantages in solving
based on multi-agent technology [37], [80]–[84]. It regards the energy management of single MG [92]. But for MMG
each MG as an agent, and each agent can spontaneously system, due to the energy management process involves many
communicate and exchange information with the neighbor- nonlinear constraints, optimization variables and interaction
ing agents according to the pre-set communication proto- with the system information, it is easy to lead dimension
col, to realize power flow and intelligent control between disaster problem, which make the optimization algorithm
MMG [81]. At the spatial scale, MMG information interac- unable to fast convergence, and then cannot use in the real-
tion based on multi-agent system adopts a three-level linkage time operation. To solve this problem, reference [93] used
hierarchical control strategy of system layer, node layer and the deep reinforcement learning algorithm to calculate the
device layer. Its multi-level target information is delivered energy optimization strategy online, and effectively improves
from top to bottom, and at the same time, the schedul- the energy efficiency of building through real-time feedback
ing information is uploaded from bottom to top, to achieve and control.
coordinated and optimized control of energy in the whole
system [82]. Reference [83], based on the MMG 3-level III. THE UNCERTAIN CHALLENGES IN MULTI-MICROGRID
energy coordination control framework of multi-agent sys- SYSTEM
tem, discussed the influence of MMG operation and energy From the MMG that have been studied so far, uncertainty
storage system operation, and proposes the solution process factors exist widely in power system [94]. In addition, MMG
of global optimization and regional autonomy. In particular, has higher renewable energy penetration, the impact of uncer-
the privacy disclosure problem caused by a large amount tainty on MMG is generally higher than that of in sin-
of information interaction in MMG is a big challenge. Ref- gle MG. Therefore, this section focuses on discussing the
erence [84] proposed a distributed scheduling consistency MMG energy management model with uncertainties, which
algorithm based on multi-agents to solve the energy manage- includes three aspects. The first one is the uncertainties in
ment problem of MMG. This method only needs informa- operation, including the output of renewable energy and the
tion exchange between adjacent agents to achieve the global uncertainty of various load demands [95]. Although MMG
optimization, which greatly reduces communication volume can make the system run stably through energy complemen-
and effectively improves the problem of user privacy disclo- tarity; it is still a system with limited capacity in essence
sure. Reference [85] proposed a multi-agent based optimal compared with large power grid. The power fluctuation of
scheduling and trading scheme to co-optimize the coupled renewable energy and load demand are more random, which
system in which multi-microgrids integrated with an urban will affect the real-time power balance of the system. The
transportation network. Reference [86] proposed an improved second is the uncertainty of the real-time electricity price
multi-agent consensus algorithm to improve the utilization caused by the reform of the electricity price system. This
rate of renewable energy resources and solve energy dispatch type of uncertainty essentially belongs to the category of the
optimization of islanded multi-microgrids. electricity market. If the demand uncertainty from users and
the market cannot be reasonably addressed, it will be difficult
3) ARTIFICIAL INTELLIGENCE TECHNOLOGY for the MG operator to make the optimal power scheduling
With the rapid development of artificial intelligence technol- plan. Thirdly, the uncertainty caused by power system faults
ogy, deep learning, enhanced learning, and other theoretical and accidents. In the initial design and planning of MMG,
methods showed great advantages in data analysis, predic- the scenarios involved in power system faults and accidents
tion, classification, and other aspects, especially in the pro- should be fully considered, to provide reasonable counter-
cessing of massive data [87], [88]. At present, many related measures when accidents occur, reduce the probability of
technologies have been applied in MMG operation, energy system collapse, and make it more robust [96].
management and other aspects [89]–[91]. Reference [89]
proposed the combination of deep neural network and rein- A. CHARACTERISTICS AND MODELING OF UNCERTAIN
forcement learning. In this model, there is no need of user’s The modeling of uncertain factors is the basis of ana-
information, and it use the historical data to train the deep lyzing the uncertainty of MMG. The cognitive degree of
neural network, which can automatically generate MMG uncertainty can be divided into randomness, fuzziness, and
power exchange strategy under given new input conditions. unknowability [94]. This paper mainly studies randomness,

VOLUME 10, 2022 77087

 G. Ma et al.: Review on Optimal Energy Management of MMG System Considering Uncertainties

which is the most widely discussed in power system analy- electricity price forecasting depends such as weather condi-
sis. The uncertainty characterization methods at randomness tions, bidding process, renewable power generation, and load
level mainly include statistical uncertainty characterization demand [110]. All these uncertainties factors make electricity
method (which can give probability distribution function) and price volatile and uncertainty.
scenario-type uncertainty characterization method (which Different authors have provided various methods for elec-
can give fluctuation interval) [97]. tricity price forecasting. 1) Statistical models; 2) Funda-
In the past decade, the modeling of uncertainty factors mental models; 3) Artificial intelligence models. Statistical
in power system has made great progress. Various statisti- models are a mathematical model which include previous
cal methods have been applied to the modeling and anal- electricity prices and exogenous variables such as weather,
ysis of renewable energy generation [98], diversified load production variables, etc. These models can use different
demand [99], energy market transaction price [29] and equip- methods to solve, such as time series models, regression mod-
ment operation state evaluation [100], all of which accurately els, and exponential smoothing methods [111]. For example,
reveal the random statistical behavior of uncertainty factors. Time series methods with seasonality and exogenous factors
for short term electricity price forecasting in the Russian
1) POWER SUPPLY AND LOAD DEMAND power market, Nord pool and PJM market [112]. Regression
Probability prediction technology is divided into interval pre- models, e.g., multiple regression models are used to learn the
diction and density prediction. The former gives the confi- relationship between dependent (interest) variable and other
dence interval of distributed renewable energy output or load independent variables. Reference [113] used double seasonal
demand at a certain confidence level, and the latter is the prob- exponential smoothing model to forecast hourly spot prices
ability density function or cumulative distribution function of from the Spanish market. Fundamental models can establish
the output at the future time [101]. Compared with interval the mathematical model of the factors affecting the price of
prediction, density prediction can provide more reference electricity, it can also use the multi-agent simulation, equi-
information. The commonly used density prediction methods librium, game theory, simulating the operation of different
include quantile regression, kernel density estimation and so agents generating units and demand load which are being
on [102]. Probability prediction technology has the obvious interacting with each other and give the price process by com-
difference from other uncertain factor modeling methods is paring the demand and supply in the desired market. Artificial
that the traditional modeling method is based on the analysis intelligence models can handle complex and non-linear prob-
of historical data, while the probabilistic prediction tries to lems. For example, the artificial neural network and artificial
obtain the variation characteristics of uncertain factors in cooperative search algorithm was applied in [114] to forecast
the short-term future, so the probabilistic prediction is more electricity price. Different deep neural network has also been
suitable for the short-term operation of power system. applied to forecast the Austrian electricity wholesale mar-
Although many literatures have carried out relevant studies ket price [115]. Reference [116] presents the hybridization
on probability prediction of wind power [103], photovoltaic of global sensitivity analysis with data-driven techniques to
power generation [104] and load demand [105], there are evaluate the Mexican electricity market interaction and assess
still several problems to be solved in probability prediction: the impact of individual parameters concerning marginal
1) Quantile crossover exists in quantile regression method, prices.
that is, quantile predicted value does not increase as the
corresponding probability value increases [106]. 2) Density 3) POWER SYSTEM RANDOM FAULTS
leakage in kernel density estimation leads to the problem that Compared with power fluctuation of distributed renewable
the predicted value exceeds the value range of the random energy and load demand, random faults of power system
variable [107]. 3) The technical bottleneck of high dimen- components such as lines, generators and transformers have
sional probability prediction considering the temporal and a more significant impact on system operation, and their
spatial correlation of power output and load demand has not uncertainties are more difficult to characterize [117]. In the
been broken [108]. With the improvement of the penetration present study, discrete probability distribution function is
of distributed renewable energy and diversified load demand generally used to represent random faults of components.
in the power system, it is more significance to overcome Reference [118] pointed out that the statistical index of com-
the above technical problems and improve the accuracy and ponent failure is difficult to accurately express the possibility
reliability of probability prediction. of line failure in actual operation, so fuzzy mathematics is
adopted to express the uncertainty of operation of overhead
2) ELECTRICITY PRICE line.
Electricity price forecasting is essential to the operation of For the modeling of power system fault components,
electricity market which has critical influence on economic a feasible idea is to establish the fuzzy set of its probabil-
operation of modern power system. Electricity market needs ity distribution based on the historical data of component
a constant balance between power supply and load demand, faults, other than to directly establish its probability
and it could not be stored at a large level due to some distribution model, and then solve the stochastic opti-
economic reasons [109]. There are various factors on which mization problem considering component faults, to deal

77088 VOLUME 10, 2022

G. Ma et al.: Review on Optimal Energy Management of MMG System Considering Uncertainties

TABLE 3. Uncertainty modeling technologies in MMGEMS.

with the influence of component faults on power system to go off the main grid under the condition of significant
operation [119]. frequency or voltage fluctuation, thus affecting the stable
Uncertainty problems and modeling methods encountered operation of the system [122], [123]. Uncertainty factors will
by mmg are summarized in TABLE 3 make the system operation stability index (such as static
voltage stability and transient voltage stability) have proba-
B. ANALYSIS OF THE INFLUENCE OF UNCERTAINTIES bility characteristics. And system operation stability needs to
1) UNCERTAINTY AFFECTING ASSESSMENT further consider the boundary of the probability distribution
Aiming at the uncertainty problems encountered in the pro- range, which can lead to the system operation in an extreme
cess of energy management of MMG, quantitative evaluation danger scenario [124].
of the influence of the above uncertainty factors becomes
the key to the uncertainty analysis of power system. The d: ECONOMY
influence of uncertain factors includes the following aspects. In the dispatching of high proportion distributed renewable
energy power system, it is necessary to consider the influ-
a: SECURITY ence of system operation uncertainties. The fluctuation of
The uncertain characteristics of distributed renewable energy diversified load demand, distributed renewable energy out-
and various load demand will directly affect the safety of put and other factors will directly affect the output alloca-
power system operation [120]. Random failure of compo- tion and reserve capacity retention of unit on-off plan, thus
nents and random fluctuation of distributed renewable energy affecting the economy of system operation. In addition, the
output will both increase the probability of line power flow electricity price will be affected by the uncertainty of supply
overload and node voltage exceeding the limit in operation, and demand balance of real-time electricity market transac-
and further affect the safety power supply in system. tions, which will further affect the economy of users’ energy
consumption [125].
b: RELIABILITY
With the improvement of renewable energy penetration, 2) MITIGATION MEASURES FOR UNCERTAINTY
and considering the uncertainty of renewable energy output, The uncertainty faced by the MMG will bring many problems
to guarantee system has plenty of power supply capacity, to the operation of the power system. To solve these oper-
the number of renewable energy generators will significantly ation problems including security, reliability, stability, and
increase [1], [2], in this situation, when it meets the extreme economy, the researchers generally consider the following
weather conditions, the energy power cuts will influence technical methods to tackle these mentioned uncertainties:
power reliability [119], [121]. In addition, the uncertainty of
users’ energy consumption behavior and transaction behavior a: ENERGY STORAGE DEVELOPMENT AND APPLICATION
will bring great challenges to the system operation, affect the Compared with other power supply and load demand, energy
real-time supply-demand balance of the system, and further storage can transfer power and energy in different time
affect the reliability of the system operation. and different space, which make its operation more flexi-
ble [126]. At present, energy storage has been applied in
c: STABILITY energy arbitrage and electric energy regulation. In addition,
As we all know, some uncertainty factors will increase the energy storage has the capacity of active and reactive power
probability of the system frequency and node voltage fluctu- coordinated scheduling, which plays an important role in
ation, and it will lead renewable energy generators are easy frequency and voltage regulation of power system, alleviating

VOLUME 10, 2022 77089

 G. Ma et al.: Review on Optimal Energy Management of MMG System Considering Uncertainties

grid congestion and improving power quality. Aiming at will coordinate the flexible resources on the user-side to par-
the challenge brought by large-scale distributed renewable ticipate in the operation of the power system through the price
energy to power system operation, energy storage can be signal, which can promote the consumption of renewable
configured to suppress the volatility generated by renewable energy, and realize the privacy protection of the user.
energy, to avoid abandon of renewable energy, and to promote
renewable energy consumption [127]. c: MULTI-ENERGY COMPLEMENT
In recent years, energy storage technology develops Through multiple energy networks interconnection and flex-
towards high energy density, high conversion efficiency and ible energy conversion, the regional integrated energy system
low cost, which promotes the widely application of energy can satisfy various energy demand and achieve clean and
storage in the power system [128]. Electrochemical energy efficient energy utilization [136]. To reduce the influence of
storage has the characteristics of flexible configuration and the uncertainty of renewable energy generation, the electric
easy operation and maintenance. In recent years, remark- energy generated by intermittent energy is converted into
able achievements have been made in theoretical research other forms of energy through different coupling mechanisms
and engineering practice. Electrochemical energy storage is during the operation of the power system.
widely used in energy storage power stations and electric Electrical energy and heat/cold can be coupled in a variety
vehicles, including lithium-ion batteries, ultracapacitors and of ways. Electric boilers, water heaters and other devices
full vanadium liquid flow battery, etc. [129]. At the same based on Joule’s law convert electrical energy directly into
time, liquid air energy storage and megawatt flywheel energy heat and store it, thereby promoting renewable energy con-
storage have been applied in micro-energy network. With the sumption [137]. The heat pump drives the compressor by
development of low-carbon transformation of energy struc- electric energy and absorbs or discharges heat energy with
ture and energy storage technology, energy storage with low the refrigerant as the carrier to meet the needs of users for
cost and high reliability will become an important regulation heating or cooling [138].
means of energy system to solve the uncertainties. In particular, the continuous development of power to Gas
technology in recent years will become the key to promote the
b: LOAD DEMAND RESPONSE AND REGULATION absorption of renewable energy and the integration of hetero-
Take advantage of users’ side load control technology to geneous energy [139]. P2G converts electrical energy into sta-
deal with the fluctuation of renewable energy has become a ble high energy density gas, such as hydrogen and methane,
research focus on the whole world [130]. Usually, the loads thus storing intermittent and fluctuating renewable energy in
in MMG can be divided into flexible and inflexible loads. the form of stable chemical energy. Gas power plants, micro-
And the flexible loads can response to price signals and gas turbines and other electrical coupling equipment have
participate in the demand response programs. Based on the good dynamic response speed, which can quickly adjust the
demand response characteristics, MMG operator can reduce output according to the renewable energy and load fluctuation
its total daily costs [131]. Reference [132] presented a smart to ensure the smooth operation of the system [140].
home energy management system that includes flexible appli- Compared with the traditional optimal scheduling of power
ances, electric vehicles, and energy storage units, this sys- system, the coordinated operation of multi-energy comple-
tem can transform the consumer into an active prosumer by mentary integrated energy system can give full attention to
demand management. In addition, some researchers studied the different time scales advantages of dispatching response
the regulation model and parameter identification technology of different energy systems and realize the complementary of
of temperature load on different time scales (hour, minute cold-heat-power-gas.
and second) [133], to solve the power fluctuation problem
of renewable energy. Similarly, in industrial users, user-side C. MULTI-MICROGRID SYSTEM DISPATCHING WITH
resources such as electrolytic aluminum and irrigation pump UNCERTAINTIES
stations can also provide auxiliary services to solve the prob- MMG’s energy management and control strategy is the basis
lem of power fluctuation of renewable energy generation. of influencing the coordination and stable operation of MGs,
Reference [134] introduced the demand response and eco- but MMG still faces the influence of uncertain factors on its
nomic energy storage dispatch to enhance self-coordination safe and economic in the process of operation [25]. Therefore,
and self-balancing ability among different resources and it is a major difficulty of MMG energy management system
established an effective modeling and optimization method to modeling to effectively deal with various uncertainties in
maximization of renewable energy utilization and minimiza- MMG. The following three types of models proposed by
tion of overall system costs in MMG. researchers for uncertain factors are mainly introduced.
With the gradual opening of the electricity market, the
rapid development of distributed renewable energy and the 1) STOCHASTIC OPTIMIZATION
experimental application of blockchain technology to support Stochastic optimization includes uncertain parameters, and
decentralized trading, the tradable energy system and mecha- we assume these uncertainties variables follow a predefined
nism for distributed market are gradually emerging [135]. The probability distribution and then we can transform it into a
transaction mechanism based on the tradable energy market deterministic mathematical programming problem and solve

77090 VOLUME 10, 2022

G. Ma et al.: Review on Optimal Energy Management of MMG System Considering Uncertainties

it [141]. To a certain extent, this method can reduce the two-stage robust optimization models, because compared
complexity of scenarios faced by MMG system in energy with static robust optimization or single-stage robust opti-
management. Reference [142] adopted the method based on mization methods, two-stage robust optimization models
random scenarios to deal with the uncertainty of renewable have greatly improved the conservatism of robust optimiza-
energy generation and used a backward scenario reduction tion [148]. Reference [149] proposed a two-layer, two-stage
method to reduce the generated scenarios and the amount robust optimization model for the uncertainty of renewable
of calculation. Reference [143] proposed an energy manage- energy output, load demand and line faults in AC-DC hybrid
ment strategy based on unexpected events for MMG sys- MMG. The column-and-constraint generation algorithm was
tem. Power operators consider the probability of unexpected used to transform the Min-Max-Min problem into a two-stage
events in energy management and adopt random optimization mixed-integer linear programming problem, which can be
strategies based on different scenarios. The results show that solved quickly and effectively. Reference [150] formulated
this model has obvious advantages in preventing economic a cost minimization robust model for MMG energy manage-
losses. ment, which considers the renewable energy output and load
There are also related literatures that use chance con- demand uncertainty, it has been transformed to a min-max
strained programming, which is also a SO. It stipulates robust counterpart. The developed model provides immu-
that the probability of the establishment of some constraints nity against the worst-case realization within the provided
should be greater than a certain confidence level, which may uncertainty bounds. At present, the main problem of robust
better meet the needs of decision makers in practical appli- optimization is to improve the conservatism of uncertain sets,
cation [27]. SO mainly use related mathematical methods to because the robust optimization ignores the probability dis-
generate random scenarios to describe the possible uncer- tribution of some available uncertain parameters, resulting in
tainty problems in power system, but the difficulty is how the uncertain sets’ description of uncertain parameters is too
to generate a small number of random scenes to represent conservative. In addition, robust optimization models usually
more uncertain scenes. Reference [144] used Monte Carlo use the worst scenario to deal with uncertainty, which often
algorithm to generate discrete random scenarios to describe leads to high conservatism and poor economy [151].
the uncertainty of relevant parameters, but Monte Carlo
method has a slow convergence rate and a large amount of 3) DISTRIBUTIONALLY ROBUST OPTIMIZATION
calculation. Reference [145] proposed a two-point estimation It is well known that the deterministic probabilistic distribu-
method instead of Monte Carlo method to be embedded into tion needed in stochastic optimization can only be observed
the gradient-based particle swarm optimization algorithm, indirectly through finite datasets and robust optimization has
to avoid the problem of too slow convergence speed and an over-conservativeness feature because it ignores proba-
reduce the computational burden. In addition, the main tech- bility distribution information. Recently, a distributionally
nical difficulty of stochastic optimization is how to accurately robust optimization (DRO) method that remedies the SO’s
describe the probability distribution of uncertain parameters. concreteness of probability information and RO’s ignorance
Compared with single microgrid system, the energy inter- of probability information was proposed in [152], [153].
action and power transaction between each microgrid and Distributionally robust optimization is a modeling pro-
the main grid will increase. Power transmission loss and gram, adopting the worst-case approach, where the worst
real-time electricity price will make the optimization oper- case is chosen from a proscribed ambiguity distribution set.
ation of MMG system more complicated [146]. All of these In many situations, the probability distribution function of
make it difficult to accurately obtain the probability distri- renewable energy, load demand and electricity price cannot
bution of uncertain parameters. What’s more, the stochastic be exactly obtained; instead, its partial information such as
optimization always faces the balance of calculation com- mean vector and covariance matrix can be collected from
plicity and solution accuracy, which is the key problem that the historical data. Unlike the SO approach that requires full
restricts the further development of the stochastic program- PDF specification, the assumption for DRO-based models is
ming model. that only partial information about the distributions of the
uncertain parameters is available.
2) ROBUST OPTIMIZATION More recently, DRO approaches have been applied to
Different from the stochastic optimization, the robust opti- model uncertainty in power system problems such as eco-
mization (RO) does not need to know the probability distri- nomic dispatch [154], optimal power flow [155], investment
bution of the uncertain parameters in advance but expresses decisions [156], and unit commitment [157], etc. Refer-
the fluctuation range of the uncertain parameters by setting ence [155] adopted a statistical inference technique to con-
up an uncertain set. If the uncertain parameters fluctuate struct ambiguity sets for the discrete distributions, which
within the given uncertain set, the RO can generally obtain is data-driven, and more data leads to the less conservative
feasible solutions [147]. Under the same conditions, the solution. Reference [157] proposed a Wasserstein metric-
computation of the model is less than that of the method based distributionally robust approximate framework, for unit
based on SO. At present, there are relatively few studies commitment problem to manage the risk from uncertain wind
on the application of RO in MMG, and most of them are power forecasted errors. From the studies that have been

VOLUME 10, 2022 77091

 G. Ma et al.: Review on Optimal Energy Management of MMG System Considering Uncertainties

TABLE 4. Method of solving MMG dispatching with uncertainties.

done so far, instead of using a certain probability distribution will show exponential growth, which is extremely time-
or constructing a deterministic uncertainty set, DRO lies all consuming [160]. The mathematical programming method
possible probability distributions in an ambiguity set, which has strict requirements on the objective function and con-
is data-driven and assumed to contain the true distribution. straint conditions, and the complexity of energy management
The best decision is made under worst- case probability of MMG limits its application. However, through appropriate
distribution so that all possible probability distributions in model simplification, the energy management problem of
ambiguity set are immunized. MMG can transform as mixed integer linear programming
The methods of solving MMG dispatching model with of multi-stage and multi-game players, and relevant mature
uncertainties are shown in Table 4. algorithms can be used to solve it. However, the simplification
of the model means the loss of some useful information,
D. SOLVING METHODS OF MULTI-MICROGRID SYSTEMS thus by using the simplification means losing the opportunity
The energy management of MMG contains not only dis- to find the global optimal solution. However, the heuristic
crete variables such as transaction strategy formulation and algorithm is usually independent of specific application prob-
transaction intention analysis of different subjects [23], lems, and the modeling method is relatively loose, which can
but also continuous variables such as power energy out- easily deal with the uncertainty of information. Therefore,
puts of different MGs [25]. Both single objective opti- it is widely used in the energy management of MMG. But
mization [34] and multi-objective optimization [143] can meanwhile, this method cannot guarantee to find the optimal
be considered. There are both linear and nonlinear con- solution, and the solving efficiency is low.
straints in the constraints [155], [158]. It is faced with an
uncertain environment, including the random fluctuation of IV. CHALLENGES AND FUTURE DIRECTIONS
renewable energy and the real-time change of diversified With more and more power electronic devices (including
load demand, as well as the volatility of electricity price distributed energy resources and loads) connected to multi-
and fuel price. Therefore, the energy management problem microgrid systems, the development of communication tech-
of MMG is essentially a problem of multi-scenes, multi- nology, the design and implementation of complex power
objectives, nonlinear and mixed uncertainty. To solve this control modules, the maturity of energy transaction market
problem, enumeration method [27], [141]–[145], mixed inte- and the emergence of emerging technologies, The smart
ger programming method [79], [159], heuristic algorithm power grid industry will face new challenges. These chal-
[88]–[91] have been studied respectively. Enumeration, as the lenges will not only affect the realization of multi-microgrid
name implies, enumerates all possible combinations of opti- energy management, but will also have different charac-
mized variables. When the number of combinations is small, teristics in MMG research fields, such as, 1) massive dis-
the method is simple and efficient, and can ensure the tributed energy resources will provide a new possibilities for
global optimal solution. However, if the number of variables MMG planning and distributed operation control [20], [59];
is large and the solution space is large, the combination 2) different energy characteristics and differences in time and

77092 VOLUME 10, 2022

G. Ma et al.: Review on Optimal Energy Management of MMG System Considering Uncertainties

space scales in the integrated energy system will significantly challenges to the future application of MMG. On the other
improve the MMG energy efficiency [136], [161], [162]; hand, MMG topological structure is changeable, especially in
3) the development of information technology such as emergency situations such as failure or faults. How to design
blockchain technology will make energy transactions more MMG topological structure with dynamic self-reconstruction
inclusive and convenient [163], [164], etc. In summary, these capability will be a key direction of future MMG planning and
technologies will bring some new development opportunities operation. In addition, the system oscillation and circulation
to different research fields such as the planning, operation, suppression in parallel with multiple converters are the basis
control, and energy transaction of MMG. of stable operation of the system. Therefore, under different
line impedance environments, the oscillation condition fail-
A. PLANNING, OPERATION AND CONTROL ure mechanism and group-controlled oscillation suppression
1) PLANNING method of multiple converters in parallel, as well as the
As for the planning of MMG system, the existing research method of circulation suppression and power equalization of
is relatively simple which mainly considering the location multiple converters in parallel with different capacities and
of distributed renewable energy, the long-term fluctuation types, are also key points for the stable operation of MMG.
of renewable energy output, the growth of load demand,
the economic and social benefits in the whole life cycle of 3) CONTROL
equipment, etc., and it is urgent to establish a comprehensive In the process of operation, MMG must face the switching of
systematic and scientific planning and design method. In the on-off grid and the recombination of regional isolated micro-
future, the main challenges of MMG system planning are as grid. The central controller or distributed controller of MMG
follows. system should meet the requirements of control strategy at
• Aiming at the planning and design of multi-micro different time scales. However, the current central controller
energy network including cold, heat and power com- or distributed controller is mainly designed for a specific
bined supply system. The multi-microgrid system with time scale. Therefore, how to reasonably design multiple
combined cooling, heat and power system can not only time scales for each controller is the key to ensure the safety
meet users’ demand for electric energy, but also meet and reliability of MMG, which is also an urgent problem
users’ demand for heat energy. In this case, the multi- to be solved for the current MMG controller. In addition,
microgrid system is a multi-micro energy network. For as the number of microgrid increases, its structure gradually
this kind of multi-microgrid with the characteristics changes from vertical to flat and from centralized control
of integrated energy network, the coupling character- to distributed control. If each microgrid individual needs to
istics between the optimal ratio of cooling, heating communicate, massive data communication will affect the
and electricity corresponding to different structures and MMG system’s work efficiency and user privacy. How to
different energy flows are the main challenging. design effective control strategy to achieve optimal control
• Coordinated planning of distribution network and effect with minimum information traffic and reduce user
multi-microgrid. As wind and photovoltaic and other privacy leakage will be an urgent problem to be considered
renewable energy are continuously connected to the in the future development of MMG system.
power grid in the form of microgrid, the planning and
expansion planning of distribution network also need B. ENERGY MANAGEMRNT
to consider the impact of multi-microgrid access. For energy management of MMG system, further research
• Economic analysis and planning of energy storage sys- can be carried out in the following three aspects.
tem. Technically speaking, the space-time translation • First, most MMG energy management only considers
characteristics of energy storage devices are of great the energy form of electricity, while future MMG sys-
significance to the planning and operation of multi- tems are more likely to exist in the form of integrated
microgrid systems. At the same time, in the market energy system of combined cold, heat and electricity.
environment, the economic analysis of the energy stor- How to solve the coupling optimization among various
age system and the impact of its life on the planning distributed systems containing cold, heat and electricity
of the multi-microgrid system need to be reasonably is an urgent problem to be solved in future MMG
demonstrated. energy management.
• Second, considering the stochastic programming
2) OPERATION model and robust optimization model has its own
MMG is a complex system which contains multi-microgrids advantages and disadvantages, especially in the real
and various distributed power generation device. With the world we can also use the method of mathematical
increasing penetration rate of renewable energy and the exten- statistics such as approximate interval probability dis-
sive use of diversified loads, the increasing uncertainty will tribution. Therefore, how to further benefit by using
lead to power imbalance between generation and load, and this type of probability distribution well to effectively
the stability of frequency and voltage will be more serious improve the conservatism of the robust optimization
than that of traditional power system, all of which will bring model is the next main work. In addition, due to the

VOLUME 10, 2022 77093

 G. Ma et al.: Review on Optimal Energy Management of MMG System Considering Uncertainties

high renewable energy penetration of MMG, robust including physical network topologies, energy trading mech-
optimization theory has limited ability to describe the anisms and various energy scheduling and control strategies.
uncertain sets. If uncertain parameters fall outside the The research works indicate that different kinds of uncer-
uncertain sets, how to deal with such risks will also be tainty elements which suffering from operation, electricity
an urgent problem to be solved in the application of price and power system faults will bring great challenges
robust optimization theory in MMG field. to MMG. Meanwhile, these uncertainties can be mitigated
• Thirdly, reinforcement learning has been widely used by energy storage technology, electricity market technology
in energy management modeling, but it can still be and complementary collaboration among different kinds of
further studied from the following aspects: training energy (cold, heat and power). Finally, various model and
corresponding reinforcement learning model methods solving method are introduced to deal with MMG energy
for typical problems by analyzing and condensed typ- management with various uncertainties, including statistic
ical energy management optimization models in dif- optimization, robust optimization and distributionally robust
ferent scenarios, including comprehensive use of the optimization. This paper summarizes the MMG architecture,
model-based reinforcement learning and model-free influence of uncertain elements, model characteristics and
reinforcement learning methods can improve the gen- solution methods, etc. It provides an integrated solution for
eralization ability of deep reinforcement learning net- MMG with uncertainties. The challenges and future direc-
work and make it more robust in energy management tions of MMG are also highlighted.
modeling. Furthermore, advanced intelligent optimiza-
tion methods are adopted to optimize the network struc- REFERENCES
ture of deep reinforcement learning and improve the [1] H. Vella, ‘‘Ten steps to net zero: As the goal to reach net zero in the U.K.
accuracy of energy management control optimization rapidly approaches, we examine Boris Johnson’s 10-point plan for a green
industrial revolution,’’ Eng. Technol., vol. 16, no. 1, pp. 20–25, Feb. 2021.
strategy. [2] Z. Wang, B. Yang, W. Wei, S. Zhu, X. Guan, and D. Sun, ‘‘Multi-
energy microgrids: Designing, operation under new business models, and
C. MARKET TRANSACTIONS engineering practices in China,’’ IEEE Electrific. Mag., vol. 9, no. 3,
pp. 75–82, Sep. 2021.
In the new generation of information and communication [3] T. Sutikno, W. Arsadiando, A. Wangsupphaphol, A. Yudhana, and
technology, blockchain technology, as a brand-new dis- M. Facta, ‘‘A review of recent advances on hybrid energy storage sys-
tributed infrastructure and accounting technology, can pro- tem for solar photovoltaics power generation,’’ IEEE Access, vol. 10,
pp. 42346–42364, 2022.
vide trust foundation for multi-microgrid collaboration in [4] L. Yan, M. Sheikholeslami, W. Gong, M. Shahidehpour, and Z. Li,
energy management with its ingenious technical design and ‘‘Architecture, control, and implementation of networked microgrids
data governance [163]. for future distribution systems,’’ J. Modern Power Syst. Clean Energy,
vol. 10, no. 2, pp. 286–299, Mar. 2022.
The energy transactions of MMG face several problems: [5] H. Zou, S. Mao, Y. Wang, F. Zhang, X. Chen, and L. Cheng, ‘‘A survey of
the high platform management cost, the low security, and energy management in interconnected multi-microgrids,’’ IEEE Access,
the untimely consumption of scattered electricity [164]. The vol. 7, pp. 72158–72169, 2019.
[6] B. Kroposki and G. Martin, ‘‘Hybrid renewable energy and microgrid
core application value of blockchain technology in the field research work at NREL,’’ in Proc. IEEE PES Gen. Meeting, Jul. 2010,
of power transaction is establishing trust and solving the trust pp. 1–4.
problem in the information world. At the same time, it has the [7] S. Ali, S. A. A. Kazmi, M. M. Malik, A. H. U. Bhatti, M. Haseeb,
S. M. R. Kazmi, and D. R. Shin, ‘‘Energy management in high RER multi-
advantages of traceability, openness, transparency, and pre- microgrid system via energy trading and storage optimization,’’ IEEE
venting single point of failure [165]. Using the above advan- Access, vol. 10, pp. 6541–6554, 2022.
tages of blockchain technology can significantly reduce the [8] Y. Sun, Z. Cai, Z. Zhang, C. Guo, G. Ma, and Y. Han, ‘‘Coordinated
energy scheduling of a distributed multi-microgrid system based on multi-
complexity of energy transactions and protect the information agent decisions,’’ Energies, vol. 13, no. 16, p. 4077, Aug. 2020.
security [164], [166], etc. In the future electricity market, [9] F. Fallahi, M. Yildirim, J. Lin, and C. Wang, ‘‘Predictive multi-
blockchain is a good solution to exert group collaboration microgrid generation maintenance: Formulation and impact on operations
& resilience,’’ IEEE Trans. Power Syst., vol. 36, no. 6, pp. 4979–4991,
and coordinate the rights and obligations of participants. Nov. 2021.
Combined with the technical characteristics of blockchain [10] C. Stevanoni, Z. De Greve, F. Vallee, and O. Deblecker, ‘‘Long-term
and the characteristics of the future development of power planning of connected industrial microgrids: A game theoretical approach
including daily Peer-to-microgrid exchanges,’’ IEEE Trans. Smart Grid,
energy market, blockchain technology has important applica- vol. 10, no. 2, pp. 2245–2256, Mar. 2019.
tion scenarios in power wholesale market, power retail mar- [11] E. Shittu and J. R. Santos, ‘‘Electricity markets and power supply
ket, distributed power generation market transaction, power resilience: An incisive review,’’ Current Sustain./Renew. Energy Rep.,
vol. 8, no. 4, pp. 189–198, Dec. 2021.
derivatives transaction, credit evaluation of market players [12] E. Shittu, A. Tibrewala, S. Kalla, and X. Wang, ‘‘Meta-analysis of the
and other aspects [167]. strategies for self-healing and resilience in power systems,’’ Adv. Appl.
Energy, vol. 4, Nov. 2021, Art. no. 100036.
[13] D. K. Mishra, M. J. Ghadi, L. Li, J. Zhang, and M. J. Hossain, ‘‘Active
V. CONCLUSION distribution system resilience quantification and enhancement through
MMG and its energy management system play a significant multi-microgrid and mobile energy storage,’’ Appl. Energy, vol. 311,
role in improving energy efficiency, power quality, reliabil- Apr. 2022, Art. no. 118665.
[14] Z. Xu, P. Yang, C. Zheng, Y. Zhang, J. Peng, and Z. Zeng, ‘‘Analysis on
ity of distribution systems. In this paper, the typical archi- the organization and development of multi-microgrids,’’ Renew. Sustain.
tectures, and functionalities of MMG has been introduced, Energy Rev., vol. 81, pp. 2204–2216, Jan. 2018.

77094 VOLUME 10, 2022

G. Ma et al.: Review on Optimal Energy Management of MMG System Considering Uncertainties

[15] L. Che and M. Shahidehpour, ‘‘Dc microgrids: Economic operation and [36] W. E. P. S. Ediriweera and N. W. A. Lidula, ‘‘Design and protection
enhancement of resilience by hierarchical control,’’ IEEE Trans. Smart of microgrid clusters: A comprehensive review,’’ AIMS Energy, vol. 10,
Grid, vol. 5, no. 5, pp. 2517–2526, Sep. 2014. no. 3, pp. 375–411, 2022.
[16] M. Shahidehpour, Z. Li, S. Bahramirad, Z. Li, and W. Tian, ‘‘Networked [37] K. Tazi, F. M. Abbou, and F. Abdi, ‘‘Multi-agent system for microgrids:
microgrids: Exploring the possibilities of the IIT-Bronzeville grid,’’ IEEE Design, optimization and performance,’’ Artif. Intell. Rev., vol. 53, no. 2,
Power Energy Mag., vol. 15, no. 4, pp. 63–71, Jul./Aug. 2017. pp. 1233–1292, Feb. 2020.
[17] M. N. Alam, S. Chakrabarti, and A. Ghosh, ‘‘Networked microgrids: [38] E. Bullich-Massagué, F. Díaz-González, M. Aragüés-Peñalba,
State-of-the-art and future perspectives,’’ IEEE Trans. Ind. Informat., F. Girbau-Llistuella, P. Olivella-Rosell, and A. Sumper, ‘‘Microgrid
vol. 15, no. 3, pp. 1238–1250, Mar. 2019. clustering architectures,’’ Appl. Energy, vol. 212, pp. 340–361, Feb. 2018.
[18] G. Ma, Z. Cai, P. Xie, P. Liu, S. Xiang, Y. Sun, C. Guo, and G. Dai, ‘‘A Bi- [39] S. A. Arefifar, M. Ordonez, and Y. Mohamed, ‘‘Energy management in
level capacity optimization of an isolated microgrid with load demand multi-microgrid systems—Development and assessment,’’ in Proc. IEEE
management considering load and renewable generation uncertainties,’’ Power Energy Soc. Gen. Meeting, Jul. 2017, pp. 910–922.
IEEE Access, vol. 7, pp. 83074–83087, 2019. [40] W. Jiang, K. Yang, J. Yang, R. Mao, N. Xue, and Z. Zhuo, ‘‘A multiagent-
[19] L. Xiong, Y. Tang, S. Mao, H. Liu, K. Meng, Z. Dong, and F. Qian, based hierarchical energy management strategy for maximization of
‘‘A two-level energy management strategy for multi-microgrid systems renewable energy consumption in interconnected multi-microgrids,’’
with interval prediction and reinforcement learning,’’ IEEE Trans. Cir- IEEE Access, vol. 7, pp. 169931–169945, 2019.
cuits Syst. I, Reg. Papers, vol. 69, no. 4, pp. 1788–1799, Apr. 2022. [41] J. Choi, S. I. Habibi, and A. Bidram, ‘‘Distributed finite-time event-
[20] S. M. Hakimi, A. Hasankhani, M. Shafie-khah, and J. P. S. Catalão, triggered frequency and voltage control of AC microgrids,’’ IEEE Trans.
‘‘Stochastic planning of a multi-microgrid considering integration of Power Syst., vol. 37, no. 3, pp. 1979–1994, May 2022.
renewable energy resources and real-time electricity market,’’ Appl. [42] A. M. Nakiganda, S. Dehghan, U. Markovic, G. Hug, and P. Aristidou,
Energy, vol. 298, Sep. 2021, Art. no. 117215. ‘‘A stochastic-robust approach for resilient microgrid investment planning
[21] I. Brahmia, J. Wang, H. Xu, H. Wang, and L. D. O. Turci, ‘‘Robust under static and transient islanding security constraints,’’ IEEE Trans.
data predictive control framework for smart multi-microgrid energy dis- Smart Grid, vol. 13, no. 3, pp. 1774–1788, May 2022.
patch considering electricity market uncertainty,’’ IEEE Access, vol. 9, [43] B. Zhao, X. Wang, D. Lin, M. Calvin, J. Morgan, R. Qin, and
pp. 32390–32404, 2021. C. Wang, ‘‘Energy management of multiple microgrids based on a sys-
[22] Y. Wang, Z. Huang, M. Shahidehpour, L. L. Lai, Z. Wang, and Q. Zhu, tem of systems architecture,’’ IEEE Trans. Power Syst., vol. 33, no. 6,
‘‘Reconfigurable distribution network for managing transactive energy pp. 6410–6421, Nov. 2018.
in a multi-microgrid system,’’ IEEE Trans. Smart Grid, vol. 11, no. 2, [44] N. Nikmehr and S. N. Ravadanegh, ‘‘Optimal power dispatch of multi-
pp. 1286–1295, Mar. 2020. microgrids at future smart distribution grids,’’ IEEE Trans. Smart Grid,
[23] A. Rajaei, S. Fattaheian-Dehkordi, M. Fotuhi-Firuzabad, and vol. 6, no. 4, pp. 1648–1657, Jul. 2015.
M. Moeini-Aghtaie, ‘‘Decentralized transactive energy management of
[45] J. Wu and X. Guan, ‘‘Coordinated multi-microgrids optimal control
multi-microgrid distribution systems based on ADMM,’’ Int. J. Electr.
algorithm for smart distribution management system,’’ IEEE Trans. Smart
Power Energy Syst., vol. 132, Nov. 2021, Art. no. 107126.
Grid, vol. 4, no. 4, pp. 2174–2181, Dec. 2013.
[24] B. Zhou, J. Zou, C. Y. Chung, H. Wang, N. Liu, N. Voropai, and D. Xu,
[46] A. Hussain, V. Bui, and H. Kim, ‘‘A resilient and privacy-preserving
‘‘Multi-microgrid energy management systems: Architecture, communi-
energy management strategy for networked microgrids,’’ IEEE Trans.
cation, and scheduling strategies,’’ J. Modern Power Syst. Clean Energy,
Smart Grid, vol. 9, no. 3, pp. 2127–2139, May 2018.
vol. 9, no. 3, pp. 463–476, 2021.
[47] S. Kampezidou, O. Vasios, and S. Meliopoulos, ‘‘Multi-microgrid archi-
[25] D. Yang, C. Zhang, C. Jiang, X. Liu, and Y. Shen, ‘‘Interval method
tecture: Optimal operation and control,’’ in Proc. North Amer. Power
based optimal scheduling of regional multi-microgrids with uncertainties
Symp. (NAPS), Sep. 2018, pp. 1–5.
of renewable energy,’’ IEEE Access, vol. 9, pp. 53292–53305, 2021.
[26] K. Thirugnanam, M. S. E. Moursi, V. Khadkikar, H. H. Zeineldin, [48] D. Bishnoi and H. Chaturvedi, ‘‘A review on emerging trends in smart
and M. Al Hosani, ‘‘Energy management of grid interconnected multi- grid energy management systems,’’ Int. J. Renew. Energy Res., vol. 11,
microgrids based on P2P energy exchange: A data driven approach,’’ no. 3, pp. 952–966, Sep. 2021.
IEEE Trans. Power Syst., vol. 36, no. 2, pp. 1546–1562, Mar. 2021. [49] A. G. Tsikalakis and N. D. Hatziargyriou, ‘‘Centralized control for opti-
[27] F. H. Aghdam, N. T. Kalantari, and B. Mohammadi-Ivatloo, ‘‘A stochastic mizing microgrids operation,’’ in Proc. IEEE Power Energy Soc. Gen.
optimal scheduling of multi-microgrid systems considering emissions: Meeting, Jun. 2011, pp. 1–8.
A chance constrained model,’’ J. Cleaner Prod., vol. 275, Dec. 2020, [50] E. J. Ng and R. A. El-Shatshat, ‘‘Multi-microgrid control systems
Art. no. 122965. (MMCS),’’ in Proc. IEEE PES General Meeting, Jul. 2010, pp. 1–6.
[28] M. Mazidi, N. Rezaei, F. J. Ardakani, M. Mohiti, and J. M. Guerrero, [51] M. Moghimi, P. Jamborsalamati, J. Hossain, S. Stegen, and J. Lu,
‘‘A hierarchical energy management system for islanded multi-microgrid ‘‘A hybrid communication platform for multi-microgrid energy manage-
clusters considering frequency security constraints,’’ Int. J. Electr. Power ment system optimization,’’ in Proc. IEEE 27th Int. Symp. Ind. Electron.
Energy Syst., vol. 121, Oct. 2020, Art. no. 106134. (ISIE), Jun. 2018, pp. 1215–1220.
[29] X. Fang, B.-M. Hodge, E. Du, C. Kang, and F. Li, ‘‘Introducing [52] M. Hirsch, M. Johnson, and D. Smith, ‘‘Communication architectures for
uncertainty components in locational marginal prices for pricing wind centralized monitoring and control of power equipment,’’ in Proc. Int.
power and load uncertainties,’’ IEEE Trans. Power Syst., vol. 34, no. 3, Telecommun. Energy Conf., 1996, pp. 76–81.
pp. 2013–2024, May 2019. [53] W. Liu, W. Gu, J. Wang, W. Yu, and X. Xi, ‘‘Game theoretic non-
[30] Q. Zhao and Y. Shen, ‘‘On topology control of power systems with cooperative distributed coordination control for multi-microgrids,’’ IEEE
uncertainties,’’ in Proc. 31st Chin. Control Conf., 2012, pp. 6943–6946. Trans. Smart Grid, vol. 9, no. 6, pp. 6986–6997, Nov. 2018.
[31] Q. P. Zheng, J. Wang, and A. L. Liu, ‘‘Stochastic optimization for [54] H. Qiu, W. Gu, Y. Xu, Z. Wu, S. Zhou, and G. Pan, ‘‘Robustly
unit Commitment—A review,’’ IEEE Trans. Power Syst., vol. 30, no. 4, multi-microgrid scheduling: Stakeholder-parallelizing distributed opti-
pp. 1913–1924, Jul. 2015. mization,’’ IEEE Trans. Sustain. Energy, vol. 11, no. 2, pp. 988–1001,
[32] J. Wang, X. Xu, H. Li, and H. Chen, ‘‘Two-stage robust optimization Apr. 2020.
of thermal-ESS units scheduling under wind uncertainty,’’ Energy Rep., [55] I. Brahmia, P. Zhao, and J. Wang, ‘‘Efficient energy management of
vol. 8, pp. 1147–1155, Aug. 2022. multi-microgrids systems based on distributed cooperative control in
[33] S. Li, W. Han, and L. Liu, ‘‘Robust optimization of the hub location autonomous mode,’’ in Proc. 9th Int. Conf. Power Energy Syst. (ICPES),
problem for fresh agricultural products with uncertain demand,’’ IEEE Dec. 2019, pp. 1–6.
Access, vol. 10, pp. 41902–41913, 2022. [56] K. Wu, Q. Li, Z. Chen, J. Lin, Y. Yi, and M. Chen, ‘‘Distributed
[34] Y. Cao, D. Li, Y. Zhang, Q. Tang, A. Khodaei, H. Zhang, and Z. Han, optimization method with weighted gradients for economic dispatch
‘‘Optimal energy management for multi-microgrid under a transac- problem of multi-microgrid systems,’’ Energy, vol. 222, May 2021,
tive energy framework with distributionally robust optimization,’’ IEEE Art. no. 119898.
Trans. Smart Grid, vol. 13, no. 1, pp. 599–612, Jan. 2022. [57] M. Mao, Y. Wang, L. Chang, and Y. Du, ‘‘Operation optimization for
[35] O. Yurdakul, F. Sivrikaya, and S. Albayrak, ‘‘A distributionally robust multi-microgrids based on centralized-decentralized hybrid hierarchi-
optimization approach to unit commitment in microgrids,’’ in Proc. IEEE cal energy management,’’ in Proc. IEEE Energy Convers. Congr. Exp.
Power Energy Soc. Gen. Meeting (PESGM), Jul. 2021, pp. 1–5. (ECCE), Oct. 2017, pp. 4813–4820.

VOLUME 10, 2022 77095

 G. Ma et al.: Review on Optimal Energy Management of MMG System Considering Uncertainties

[58] V. Bui, A. Hussain, and H.-M. Kim, ‘‘A multiagent-based hierarchical [78] S. K. Sahoo, N. K. Kishore, and A. K. Sinha, ‘‘Decentralized control and
energy management strategy for multi-microgrids considering adjustable fault ride-through of a multi-microgrid system,’’ IET Smart Grid, vol. 2,
power and demand response,’’ IEEE Trans. Smart Grid, vol. 9, no. 2, no. 3, pp. 464–476, Sep. 2019.
pp. 1323–1333, Mar. 2018. [79] M. Ghadimi and S.-M. Moghaddas-Tafreshi, ‘‘Interaction model in a
[59] S. Wenzhi, H. Zhang, M.-L. Tseng, Z. Weipeng, and L. Xinyang, ‘‘Hier- resilient standalone multi-microgrid system,’’ in Proc. 7th Int. Istanbul
archical energy optimization management of active distribution net- Smart Grids Cities Congr. Fair (ICSG), Apr. 2019, pp. 56–60.
work with multi-microgrid system,’’ J. Ind. Prod. Eng., vol. 39, no. 3, [80] Y. Han, K. Zhang, H. Li, E. A. A. Coelho, and J. M. Guerrero, ‘‘MAS-
pp. 210–229, Apr. 2022. based distributed coordinated control and optimization in microgrid and
[60] M. Mirzaei, R. Keypour, M. Savaghebi, and K. Golalipour, ‘‘Proba- microgrid clusters: A comprehensive overview,’’ IEEE Trans. Power
bilistic optimal Bi-level scheduling of a multi-microgrid system with Electron., vol. 33, no. 8, pp. 6488–6508, Aug. 2018.
electric vehicles,’’ J. Electr. Eng. Technol., vol. 15, no. 6, pp. 2421–2436, [81] M. M. Malik, S. A. A. Kazmi, H. W. Asim, A. B. Ahmed, and D. R. Shin,
Nov. 2020. ‘‘An intelligent multi-stage optimization approach for community based
[61] P. Tianjiao, L. Kewen, and L. Ye, ‘‘Multi-agent system based simulation micro-grid within multi-microgrid paradigm,’’ IEEE Access, vol. 8,
verification for autonomy-cooperative optimization control on active dis- pp. 177228–177244, 2020.
tribution network,’’ in Proc. Chin. Soc. Elect. Eng., vol. 35, no. 8, 2015, [82] A. Qian, L. Siyuan, W. Renbo, and F. Songli, ‘‘Research and prospect of
pp. 1864–1874. multi-agent system in energy internet system,’’ High Voltage Eng., vol. 42,
[62] A. Ouammi, H. Dagdougui, and R. Sacile, ‘‘Optimal control of power no. 9, pp. 2697–2706, 2016.
flows and energy local storages in a network of microgrids modeled as a [83] W. Yao, L. Huanhuan, and J. Liwei, ‘‘Coordinated energy coupling con-
system of systems,’’ IEEE Trans. Control Syst. Technol., vol. 23, no. 1, trol strategy and simulation analysis of microgrid cluster for intelligent
pp. 128–138, Jan. 2015. scheduling,’’ Power Syst. Technol., vol. 42, no. 7, pp. 2232–2239, 2018.
[63] X. Dou, P. Xu, Q. Hu, W. Sheng, X. Quan, and Z. Wu, ‘‘A dis- [84] B. Huang, Y. Li, H. Zhang, and Q. Sun, ‘‘Distributed optimal co-
tributed voltage control strategy for multi-microgrid active distribution multi-microgrids energy management for energy internet,’’ IEEE/CAA
networks considering economy and response speed,’’ IEEE Access, vol. 6, J. Autom. Sinica, vol. 3, no. 4, pp. 357–364, Oct. 2016.
pp. 31259–31268, 2018.
[85] Y. Liu, Y. Wang, Y. Li, H. B. Gooi, and H. Xin, ‘‘Multi-agent based
[64] S. A. Arefifar, M. Ordonez, and Y. A.-R.-I. Mohamed, ‘‘Voltage and optimal scheduling and trading for multi-microgrids integrated with
current controllability in multi-microgrid smart distribution systems,’’ urban transportation networks,’’ IEEE Trans. Power Syst., vol. 36, no. 3,
IEEE Trans. Smart Grid, vol. 9, no. 2, pp. 817–826, Mar. 2018. pp. 2197–2210, 2021.
[65] Y. Liu, H. B. Gooi, Y. Li, H. Xin, and J. Ye, ‘‘A secure distributed transac-
[86] K. Yang, C. Li, X. Jing, Z. Zhu, Y. Wang, H. Ma, and Y. Zhang,
tive energy management scheme for multiple interconnected microgrids
‘‘Energy dispatch optimization of islanded multi-microgrids based on
considering misbehaviors,’’ IEEE Trans. Smart Grid, vol. 10, no. 6,
symbiotic organisms search and improved multi-agent consensus algo-
pp. 5975–5986, Nov. 2019.
rithm,’’ Energy, vol. 239, Jan. 2022, Art. no. 122105.
[66] Y. Du and F. Li, ‘‘A hierarchical real-time balancing market considering
[87] A. Zainab, A. Ghrayeb, D. Syed, H. Abu-Rub, S. S. Refaat, and
multi-microgrids with distributed sustainable resources,’’ IEEE Trans.
O. Bouhali, ‘‘Big data management in smart grids: Technologies and
Sustain. Energy, vol. 11, no. 1, pp. 72–83, Jan. 2020.
challenges,’’ IEEE Access, vol. 9, pp. 73046–73059, 2021.
[67] S. Marzal, R. Salas, R. González-Medina, G. Garcerá, and E. Figueres,
[88] M. Massaoudi, H. Abu-Rub, S. S. Refaat, I. Chihi, and F. S. Oueslati,
‘‘Current challenges and future trends in the field of communication
‘‘Deep learning in smart grid technology: A review of recent advance-
architectures for microgrids,’’ Renew. Sustain. Energy Rev., vol. 82, no. 3,
ments and future prospects,’’ IEEE Access, vol. 9, pp. 54558–54578,
pp. 3610–3622, 2018.
2021.
[68] M. Moghimi, C. Bennett, D. Leskarac, S. Stegen, and J. Lu, ‘‘Com-
munication architecture and data acquisition for experimental Microgrid [89] Y. Du and F. Li, ‘‘Intelligent multi-microgrid energy management based
installations,’’ in Proc. IEEE PES Asia–Pacific Power Energy Eng. Conf. on deep neural network and model-free reinforcement learning,’’ IEEE
(APPEEC), Nov. 2015, pp. 1–5. Trans. Smart Grid, vol. 11, no. 2, pp. 1066–1076, Mar. 2020.
[69] N. Wehbring, J. Saat, and A. Moser, ‘‘Three-level control architecture for [90] U. R. Vinjamuri and L. R. Burthi, ‘‘Internet of Things platform for
hybrid AC/DC distribution grids,’’ in Proc. 5th IEEE Workshop Electron. energy management in multi-microgrid system to enhance power quality:
Grid (eGRID), Nov. 2020, pp. 1–8. ARBFNOCS technique,’’ Int. J. Numer. Model., Electron. Netw., Devices
Fields, vol. 35, no. 1, p. e2926, Jan. 2022.
[70] U. Manandhar, B. Wang, X. Zhang, G. H. Beng, Y. Liu, and A. Ukil,
‘‘Joint control of three-level DC–DC converter interfaced hybrid energy [91] L. Yin, T. Yu, B. Yang, and X. Zhang, ‘‘Adaptive deep dynamic pro-
storage system in DC microgrids,’’ IEEE Trans. Energy Convers., vol. 34, gramming for integrated frequency control of multi-area multi-microgrid
no. 4, pp. 2248–2257, Dec. 2019. systems,’’ Neurocomputing, vol. 344, pp. 49–60, Jun. 2019.
[71] T. Wang, Q. Long, X. Gu, and W. Chai, ‘‘Information flow modeling [92] M. F. Zia, E. Elbouchikhi, and M. Benbouzid, ‘‘Microgrids energy
and performance evaluation of communication networks serving power management systems: A critical review on methods, solutions, and
grids,’’ IEEE Access, vol. 8, pp. 13735–13747, 2020. prospects,’’ Appl. Energy, vol. 222, pp. 1033–1055, Jul. 2018.
[72] R. Aboli, M. Ramezani, and H. Falaghi, ‘‘A hybrid robust distributed [93] E. Mocanu, D. C. Mocanu, P. H. Nguyen, A. Liotta, M. E. Webber,
model for short-term operation of multi-microgrid distribution net- M. Gibescu, and J. G. Slootweg, ‘‘On-line building energy optimization
works,’’ Electric Power Syst. Res., vol. 177, Dec. 2019, Art. no. 106011. using deep reinforcement learning,’’ IEEE Trans. Smart Grid, vol. 10,
[73] J. Lin, W. Yu, N. Zhang, X. Yang, H. Zhang, and W. Zhao, ‘‘A Sur- no. 4, pp. 3698–3708, Jul. 2019.
vey on Internet of Things: Architecture, enabling technologies, security [94] Y. Li, J. Chen, and L. Feng, ‘‘Dealing with uncertainty: A survey of
and privacy, and applications,’’ IEEE Internet Things J., vol. 4, no. 5, theories and practices,’’ IEEE Trans. Knowl. Data Eng., vol. 25, no. 11,
pp. 1125–1142, Oct. 2017. pp. 2463–2482, Nov. 2013.
[74] H. Gu, ‘‘Application of integrated EMS of dispatching—Centralized [95] W. Yi, Y. Zhang, Z. Zhao, and Y. Huang, ‘‘Multiobjective robust schedul-
control in Guangzhou electric network control center,’’ in Proc. China ing for smart distribution grids: Considering renewable energy and
Int. Conf. Electr. Distrib., Dec. 2008, pp. 1–5. demand response uncertainty,’’ IEEE Access, vol. 6, pp. 45715–45724,
[75] Y. Du, F. Li, X. Kou, and W. Pei, ‘‘Coordinating multi-microgrid opera- 2018.
tion within distribution system: A cooperative game approach,’’ in Proc. [96] A. Hoballah, D.-E.-A. Mansour, and I. B. M. Taha, ‘‘Hybrid grey
IEEE Power Energy Soc. Gen. Meeting, Jul. 2017, pp. 1–5. wolf optimizer for transformer fault diagnosis using dissolved gases
[76] Y. Li, T. Zhao, P. Wang, H. B. Gooi, L. Wu, Y. Liu, and J. Ye, ‘‘Opti- considering uncertainty in measurements,’’ IEEE Access, vol. 8,
mal operation of multimicrogrids via cooperative energy and reserve pp. 139176–139187, 2020.
scheduling,’’ IEEE Trans. Ind. Informat., vol. 14, no. 8, pp. 3459–3468, [97] S. Luo and X. Guo, ‘‘Multi-objective optimization of multi-microgrid
Aug. 2018. power dispatch under uncertainties using interval optimization,’’ J. Ind.
[77] Y. Zhang, T. Zhang, R. Wang, Y. Liu, and B. Guo, ‘‘An innovative Manage. Optim., Early access, Dec. 2021, doi: 10.3934/jimo.2021208.
real-time price based distributed optimal energy management of multi- [98] S. Sun, C. Wang, Y. Wang, X. Zhu, and H. Lu, ‘‘Multi-objective optimiza-
microgrids in a smart distribution system,’’ in Proc. IEEE Innov. Smart tion dispatching of a micro-grid considering uncertainty in wind power
Grid Technol. Asia (ISGT-Asia), Nov. 2016, pp. 341–346. forecasting,’’ Energy Rep., vol. 8, pp. 2859–2874, Nov. 2022.

77096 VOLUME 10, 2022

G. Ma et al.: Review on Optimal Energy Management of MMG System Considering Uncertainties

[99] R. Koubaa, S. Bacha, and L. Krichen, ‘‘Robust energy management under [120] W. Dai, Z. Yang, J. Yu, K. Zhao, S. Wen, W. Lin, and W. Li, ‘‘Security
demand uncertainty for a hybrid FC/UC electric vehicle,’’ in Proc. 16th region of renewable energy integration: Characterization and flexibility,’’
Int. Multi-Conf. Syst., Signals Devices (SSD), 2019, pp. 50–54. Energy, vol. 187, Nov. 2019, Art. no. 115975.
[100] B. Hu, C. Pan, C. Shao, K. Xie, T. Niu, C. Li, and L. Peng, ‘‘Decision- [121] S. Huadong, X. Tao, G. Qiang, L. Yalou, L. Weifang, Y. Jun, and
dependent uncertainty modeling in power system operational reliability L. Wenfeng, ‘‘Analysis on blackout in great Britain power grid on August
evaluations,’’ IEEE Trans. Power Syst., vol. 36, no. 6, pp. 5708–5721, 9th, 2019, and its enlightenment to power grid in China,’’ in Proc. CSEE,
Nov. 2021. vol. 39, no. 21, 2019, pp. 6183–6191.
[101] H. Panamtash, S. Mahdavi, and Q. Zhou, ‘‘Probabilistic solar power [122] X. Chen, W. Du, and H. F. Wang, ‘‘Power system angular stability as
forecasting: A review and comparison,’’ in Proc. 52nd North Amer. Power affected by the reduced inertia due to wind displacing synchronous gener-
Symp. (NAPS), 2021, pp. 1–6. ators,’’ in Proc. 2nd Int. Conf. Power Renew. Energy (ICPRE), Sep. 2017,
[102] W. Can and S. Yonghua, ‘‘Theories, methodologies, and applications pp. 402–406.
of probabilistic forecasting for power systems with renewable energy [123] I. A. Hiskens and J. Alseddiqui, ‘‘Sensitivity, approximation, and uncer-
sources,’’ Autom. Electr. Power Syst., vol. 45, no. 1, pp. 2–16, 2021. tainty in power system dynamic simulation,’’ IEEE Trans. Power Syst.,
[103] C. Wan, J. Lin, J. Wang, Y. Song, and Z. Y. Dong, ‘‘Direct quantile vol. 21, no. 4, pp. 1808–1820, Nov. 2006.
regression for nonparametric probabilistic forecasting of wind power [124] S. Bu, W. Du, H. Wang, Z. Chen, L. Xiao, and H. Li, ‘‘Probabilistic
generation,’’ IEEE Trans. Power Syst., vol. 32, no. 4, pp. 2767–2778, analysis of small-signal stability of large-scale power systems as affected
Jul. 2017. by penetration of wind generation,’’ IEEE Trans. Power Syst., vol. 27,
[104] Q. Huang and S. Wei, ‘‘Improved quantile convolutional neural net- no. 2, pp. 762–770, May 2012.
work with two-stage training for daily-ahead probabilistic forecasting [125] Z. X. Pi, X. H. Li, Y. M. Ding, M. Zhao, and Z. X. Liu, ‘‘Demand response
of photovoltaic power,’’ Energy Convers. Manage., vol. 220, Sep. 2020, scheduling algorithm of the economic energy consumption in buildings
Art. no. 113085. for considering comfortable working time and user target price,’’ Energy
[105] A. Bracale, P. Caramia, P. De Falco, and T. Hong, ‘‘Multivariate quantile Buildings, vol. 250, Nov. 2021, Art. no. 111252.
regression for short-term probabilistic load forecasting,’’ IEEE Trans. [126] Z. Shi, W. Wang, Y. Huang, P. Li, and L. Dong, ‘‘Simultaneous optimiza-
Power Syst., vol. 35, no. 1, pp. 628–638, Jan. 2020. tion of renewable energy and energy storage capacity with the hierar-
[106] L. A. Fernandez-Jimenez, S. Terreros-Olarte, M. Mendoza-Villena, chical control,’’ CSEE J. Power Energy Syst., vol. 8, no. 1, pp. 95–104,
E. Garcia-Garrido, E. Zorzano-Alba, P. M. Lara-Santillan, Jan. 2022.
P. J. Zorzano-Santamaria, and A. Falces, ‘‘Day-ahead probabilistic [127] S. Xuewei, S. Xuefang, D. Wenqi, Z. Peng, J. Hongyan, W. Jinfang, and
photovoltaic power forecasting models based on quantile regression W. Yang, ‘‘Research on energy storage configuration method based on
neural networks,’’ in Proc. Eur. Conf. Electr. Eng. Comput. Sci. (EECS), wind and solar volatility,’’ in Proc. 10th Int. Conf. Power Energy Syst.
Nov. 2017, pp. 289–294. (ICPES), Dec. 2020, pp. 464–468.
[107] H. Zhang, Y. Liu, J. Yan, S. Han, L. Li, and Q. Long, ‘‘Improved deep [128] X. Zhou, Z. Fan, Y. Ma, and Z. Gao, ‘‘Research review on electrical
mixture density network for regional wind power probabilistic forecast- energy storage technology,’’ in Proc. 36th Chin. Control Conf. (CCC),
ing,’’ IEEE Trans. Power Syst., vol. 35, no. 4, pp. 2549–2560, Jul. 2020. Jul. 2017, pp. 10674–10678.
[108] S. Zhou, Y. Zhao, W. Gu, Z. Wu, Y. Li, Z. Qian, and Y. Ji, ‘‘Robust [129] G. A. Gómez-Ramírez, C. Meza, and S. Morales-Hernández, ‘‘Oppor-
energy management in active distribution systems considering temporal tunities and challenges to integrate electrochemical storage in central
and spatial correlation,’’ IEEE Access, vol. 7, pp. 153635–153649, 2019. American power grids,’’ Tecnologia en Marcha, vol. 34, no. 3, pp. 70–82,
[109] A. K. Prajapati, S. K. Srivastava, and A. Narain, ‘‘Electricity price fore- Jun. 2021.
casting: A bibliographical review,’’ in Proc. Int. Conf. Electr. Electron. [130] S. Mohseni, A. C. Brent, S. Kelly, and W. N. Browne, ‘‘Demand
Eng. (ICE), 2020, pp. 345–349. response-integrated investment and operational planning of renewable
[110] P. Wang, J. Tang, Z. Zhang, Y. Li, H. Wu, C. Ji, Z. Lin, J. Ma, H. Qi, and sustainable energy systems considering forecast uncertainties: A sys-
W. Zhang, L. Yang, and Z. Lin, ‘‘Bidding strategy optimization for power tematic review,’’ Renew. Sustain. Energy Rev., vol. 158, Apr. 2022,
generation company in carbon emission rights and electricity market,’’ Art. no. 112095.
Energy Rep., vol. 8, pp. 325–331, Aug. 2022. [131] H. Karimi and S. Jadid, ‘‘Optimal energy management for multi-
[111] I. Shah, H. Bibi, S. Ali, L. Wang, and Z. Yue, ‘‘Forecasting one-day- microgrid considering demand response programs: A stochastic multi-
ahead electricity prices for Italian electricity market using parametric and objective framework,’’ Energy, vol. 195, Mar. 2020, Art. no. 116992.
nonparametric approaches,’’ IEEE Access, vol. 8, pp. 123104–123113, [132] N. Koltsaklis, I. Panapakidis, G. Christoforidis, and J. Knápek, ‘‘Smart
2020. home energy management processes support through machine learning
[112] D. O. Afanasyev and E. A. Fedorova, ‘‘On the impact of outlier filtering algorithms,’’ Energy Rep., vol. 8, pp. 1–6, 2022.
on the electricity price forecasting accuracy,’’ Appl. Energy, vol. 236, [133] M. Song, C. Gao, M. Shahidehpour, Z. Li, S. Lu, and G. Lin, ‘‘Multi-time-
pp. 196–210, Feb. 2019. scale modeling and parameter estimation of TCLs for smoothing out wind
[113] A. Cruz, A. Muñoz, J. L. Zamora, and R. Espínola, ‘‘The effect of power generation variability,’’ IEEE Trans. Sustain. Energy, vol. 10, no. 1,
wind generation and weekday on Spanish electricity spot price fore- pp. 105–118, Jan. 2019.
casting,’’ Electric Power Syst. Res., vol. 81, no. 10, pp. 1924–1935, [134] Y. Gao and Q. Ai, ‘‘Demand-side response strategy of multi-microgrids
Oct. 2011. based on an improved co-evolution algorithm,’’ CSEE J. Power Energy
[114] A. Pourdaryaei, H. Mokhlis, H. A. Illias, S. H. A. Kaboli, S. Ahmad, and Syst., vol. 7, no. 5, pp. 903–910, Sep. 2021.
S. P. Ang, ‘‘Hybrid ANN and artificial cooperative search algorithm to [135] S. Li, W. Zhang, J. Lian, and K. Kalsi, ‘‘Market-based coordination of
forecast short-term electricity price in DE-regulated electricity market,’’ thermostatically controlled loads—Part I: A mechanism design formula-
IEEE Access, vol. 7, pp. 125369–125386, 2019. tion,’’ IEEE Trans. Power Syst., vol. 31, no. 2, pp. 1170–1178, Mar. 2016.
[115] L. Lehr and F. Valdes, ‘‘Forecasting electricity prices from European [136] B. Miao, J. Lin, H. Li, C. Liu, B. Li, X. Zhu, and J. Yang, ‘‘Day-ahead
single day-ahead coupling using artificial neural networks,’’ in Proc. energy trading strategy of regional integrated energy system considering
IEEE Electr. Power Energy Conf. (EPEC), Oct. 2021, pp. 202–207. energy cascade utilization,’’ IEEE Access, vol. 8, pp. 138021–138035,
[116] E. C. May, A. Bassam, L. J. Ricalde, M. A. E. Soberanis, O. Oubram, 2020.
O. M. Tzuc, A. Y. Alanis, and A. Livas-García, ‘‘Global sensitivity [137] J. Hao, Q. Chen, K. He, L. Chen, Y. Dai, F. Xu, and Y. Min, ‘‘A heat
analysis for a real-time electricity market forecast by a machine learning current model for heat transfer/storage systems and its application in
approach: A case study of Mexico,’’ Int. J. Electr. Power Energy Syst., integrated analysis and optimization with power systems,’’ IEEE Trans.
vol. 135, Feb. 2022, Art. no. 107505. Sustain. Energy, vol. 11, no. 1, pp. 175–184, Jan. 2020.
[117] J. Wang, H. Peng, W. Yu, J. Ming, M. J. Pérez-Jiménez, C. Tao, and [138] F. Fang, Q. H. Wang, and Y. Shi, ‘‘A novel optimal operational strategy
X. Huang, ‘‘Interval-valued fuzzy spiking neural p systems for fault for the CCHP system based on two operating modes,’’ IEEE Trans. Power
diagnosis of power transmission networks,’’ Eng. Appl. Artif. Intell., Syst., vol. 27, no. 2, pp. 1032–1041, May 2012.
vol. 82, pp. 102–109, Jun. 2019. [139] Y.-G. Son, B.-C. Oh, M. A. Acquah, R. Fan, D.-M. Kim, and S.-Y. Kim,
[118] S. Babaei, R. Jiang, and C. Zhao, ‘‘Distributionally robust distribution ‘‘Multi energy system with an associated energy hub: A review,’’ IEEE
network configuration under random contingency,’’ IEEE Trans. Power Access, vol. 9, pp. 127753–127766, 2021.
Syst., vol. 35, no. 5, pp. 3332–3341, Sep. 2020. [140] V. Mohan, R. Suresh, J. G. Singh, W. Ongsakul, and N. Madhu,
[119] A. Arab, A. Khodaei, S. K. Khator, K. Ding, V. A. Emesih, and Z. Han, ‘‘Microgrid energy management combining sensitivities, interval and
‘‘Stochastic pre-hurricane restoration planning for electric power systems probabilistic uncertainties of renewable generation and loads,’’ IEEE
infrastructure,’’ IEEE Trans. Smart Grid, vol. 6, no. 2, pp. 1046–1054, J. Emerg. Sel. Topics Circuits Syst., vol. 7, no. 2, pp. 262–270,
Mar. 2015. Jun. 2017.
VOLUME 10, 2022 77097
 G. Ma et al.: Review on Optimal Energy Management of MMG System Considering Uncertainties

[141] J. Dupačovä, N. Gröwe-Kuska, and W. Römisch, ‘‘Scenario reduction in [164] Y. Zhang, Q. Wu, and W. Hu, ‘‘A multi-microgrid thermal game model
stochastic programming,’’ Math. Program., vol. 95, no. 3, pp. 493–511, based on quantum blockchain,’’ Int. J. Heat Technol., vol. 39, no. 5,
2003. pp. 1491–1500, Oct. 2021.
[142] Z. Lyu, L. Zheng, and X. Yang, ‘‘Dual-layer optimized grid-connected [165] M. Tao, Z. Wang, and S. Qu, ‘‘Research on multi-microgrids scheduling
operation strategy of electro-thermal multi-microgrid system considering strategy considering dynamic electricity price based on blockchain,’’
the uncertainty of renewable energy sources,’’ J. Renew. Sustain. Energy, IEEE Access, vol. 9, pp. 52825–52838, 2021.
vol. 13, no. 6, Nov. 2021, Art. no. 065503. [166] M. D. Ghotbabadi, S. D. Dehnavi, H. Fotoohabadi, H. Mehrjerdi, and
[143] X. Hu and T. Liu, ‘‘Co-optimisation for distribution networks with H. Chabok, ‘‘Optimal operation and management of multi-microgrids
multi-microgrids based on a two-stage optimisation model with dynamic using blockchain technology,’’ IET Renew. Power Gener., early access,
electricity pricing,’’ IET Gener., Transmiss. Distrib., vol. 11, no. 9, 2021, doi: 10.1049/rpg2.12238.
pp. 2251–2259, Jun. 2017. [167] Z. Bao, Q. Wang, W. Shi, L. Wang, H. Lei, and B. Chen, ‘‘When
[144] N. Bazmohammadi, A. Tahsiri, A. Anvari-Moghaddam, and blockchain meets SGX: An overview, challenges, and open issues,’’ IEEE
J. M. Guerrero, ‘‘Optimal operation management of a regional network of Access, vol. 8, pp. 170404–170420, 2020.
microgrids based on chance-constrained model predictive control,’’ IET
Gener., Transmiss. Distrib., vol. 12, no. 15, pp. 3772–3779, Aug. 2018.
[145] Y. Huang, L. Wang, W. Guo, Q. Kang, and Q. Wu, ‘‘Chance constrained GUOLONG MA received the bachelor’s and
optimization in a home energy management system,’’ IEEE Trans. Smart
master’s degrees in electrical engineering from
Grid., vol. 9, no. 1, pp. 252–260, Jan. 2018.
[146] X. Min, X. Ji, X. Hu, P. Cheng, Y. Du, and M. Liu, ‘‘Autonomous the South China University of Technology,
optimized economic dispatch of active distribution system with multi- Guangzhou, China, in 2017 and 2020, respectively.
microgrids,’’ Energy, vol. 153, pp. 479–489, Jun. 2018. He is currently pursuing the Ph.D. degree with
[147] B. Tan, H. Chen, X. Zheng, and J. Huang, ‘‘Two-stage robust optimization the University of Birmingham, Birmingham, U.K.
dispatch for multiple microgrids with electric vehicle loads based on a His current research interests include renewable
novel data-driven uncertainty set,’’ Int. J. Electr. Power Energy Syst., energy integration and distribution system opti-
vol. 134, Jan. 2022, Art. no. 107359. mization and scheduling.
[148] D. Bertsimas and M. Sim, ‘‘The price of robustness,’’ Oper. Res., vol. 52,
no. 1, pp. 35–53, Jan. 2004.
[149] H. Qiu, B. Zhao, W. Gu, and R. Bo, ‘‘Bi-level two-stage robust optimal
scheduling for AC/DC hybrid multi-microgrids,’’ IEEE Trans. Smart JIANING LI (Member, IEEE) received the B.Eng.
Grid, vol. 9, no. 5, pp. 5455–5466, Sep. 2018.
[150] A. Hussain, V. H. Bui, and H. M. Kim, ‘‘Robust optimization-based degree in electrical and electronic engineering
scheduling of multi-microgrids considering uncertainties,’’ Energies, from the Huazhong University of Science and
vol. 9, no. 4, p. 278, 2016. Technology, Wuhan, China, and from the Univer-
[151] G. Zhu, J. Lin, Z. Luo, S. Dai, L. Qin, and C. Liu, ‘‘Review of robust sity of Birmingham, Birmingham, U.K., respec-
optimization for generation scheduling in power systems,’’ Proc. CSEE, tively, in 2011, and the Ph.D. degree from the
vol. 37, no. 20, pp. 5881–5892, 2017. School of Electronic, Electrical and Systems Engi-
[152] L. Yang, Y. Xu, H. Sun, and W. Wu, ‘‘Tractable convex approxima- neering, University of Birmingham, in 2016.
tions for distributionally robust joint chance-constrained optimal power He was an Assistant Professor in smart energy sys-
flow under uncertainty,’’ IEEE Trans. Power Syst., vol. 37, no. 3, tems with the University of Birmingham. He then
pp. 1927–1941, May 2022. joined WSP as a Principal Power Systems Engineer, in 2022. His research
[153] A. Zhou, H. Zhai, M. Yang, and Y. Lin, ‘‘Three-phase unbalanced
interests include distributed energy resource, energy storage systems, electric
distribution network dynamic reconfiguration: A distributionally robust
approach,’’ IEEE Trans. Smart Grid, vol. 13, no. 3, pp. 2063–2074, vehicle, energy management systems, and smart grid technology. He is
May 2022. currently the Chair of IEEE U.K., and Ireland Power and Energy Society.
[154] W. Wei, F. Liu, and S. Mei, ‘‘Distributionally robust co-optimization of
energy and reserve dispatch,’’ IEEE Trans. Sustain. Energy, vol. 7, no. 1,
pp. 289–300, Jan. 2016. XIAO-PING ZHANG (Fellow, IEEE) received
[155] W. Huang, W. Zheng, and D. J. Hill, ‘‘Distributionally robust optimal the B.Eng., M.Sc., and Ph.D. degrees in electri-
power flow in multi-microgrids with decomposition and guaranteed con- cal engineering from Southeast University, China,
vergence,’’ IEEE Trans. Smart Grid, vol. 12, no. 1, pp. 43–55, Jan. 2021. in 1988, 1990, and 1993, respectively.
[156] F. Alismail, P. Xiong, and C. Singh, ‘‘Optimal wind farm allocation He was an Associate Professor with the Univer-
in multi-area power systems using distributionally robust optimiza- sity of Warwick, U.K. From 1993 to 1998, he was
tion approach,’’ IEEE Trans. Power Syst., vol. 33, no. 1, pp. 536–544,
with the China State Grid EPRI (NARI Group)
Jan. 2018.
[157] R. Zhu, H. Wei, and X. Bai, ‘‘Wasserstein metric based distributionally on EMS/DMS advanced application software
robust approximate framework for unit commitment,’’ IEEE Trans. Power research and development. From 1998 to 1999,
Syst., vol. 34, no. 4, pp. 2991–3001, Jul. 2019. he was visiting UMIST. From 1999 to 2000, he was
[158] M. Yan, M. Shahidehpour, A. Paaso, L. Zhang, A. Alabdulwahab, and an Alexander-von-Humboldt Research Fellow with the University of Dort-
A. Abusorrah, ‘‘Distribution network-constrained optimization of Peer- mund, Germany. He is currently a Professor of electrical power systems with
to-Peer transactive energy trading among multi-microgrids,’’ IEEE Trans. the University of Birmingham, U.K. He is also the Director of the Smart Grid,
Smart Grid, vol. 12, no. 2, pp. 1033–1047, Mar. 2021. Birmingham Energy Institute. He has coauthored the first and second edition
[159] M. Ott, M. AlMuhaini, and M. Khalid, ‘‘A MILP-based restoration tech- of the monograph Flexible AC Transmission Systems: Modelling and Control
nique for multi-microgrid distribution systems,’’ IEEE Access, vol. 7, (Springer, in 2006 and 2012). He has coauthored the book Restructured
pp. 136801–136811, 2019. Electric Power Systems: Analysis of Electricity Markets with Equilibrium
[160] G. L. Valiant, ‘‘The complexity of enumeration and reliability problems,’’ Models (IEEE Press/Wiley, in 2010). His research interests include modeling
SIAM J. Comput., vol. 8, no. 3, p. 410, 1979. and control of HVDC, FACTS and renewable energy, distributed generation
[161] S. Zhang, W. Gu, S. Yao, S. Lu, S. Zhou, and Z. Wu, ‘‘Partitional control, energy market operations, and power system planning.
decoupling method for fast calculation of energy flow in a large-scale heat
Dr. Zhang has been made a fellow of IEEE for contributions to modeling
and electricity integrated energy system,’’ IEEE Trans. Sustain. Energy,
and control of high-voltage dc and ac transmission systems. He is an IEEE
vol. 12, no. 1, pp. 501–513, Jan. 2021.
[162] X. Zhong, W. Zhong, Y. Liu, C. Yang, and S. Xie, ‘‘Optimal energy man- PES Distinguished Lecturer on HVDC, FACTS, and wave energy generation.
agement for multi-energy multi-microgrid networks considering carbon He is the Chair of the IEEE WG on Test Systems for Economic Analysis.
emission limitations,’’ Energy, vol. 246, May 2022, Art. no. 123428. He is an Advisor to the IEEE PES U.K. and Ireland Chapter. He has
[163] Z. Zhao, J. Guo, X. Luo, J. Xue, C. S. Lai, Z. Xu, and L. L. Lai, been appointed to the Expert Advisory Group, U.K. Government’s Offshore
‘‘Energy transaction for multi-microgrids and internal microgrid based Transmission Network Review, since 2020.
on blockchain,’’ IEEE Access, vol. 8, pp. 144362–144372, 2020.

77098 VOLUME 10, 2022