Professional Documents
Culture Documents
Good Layer3
Good Layer3
Adj
Pk,i,t Adjustable power of unit i in MG k at t. research challenges by [11]–[13] and a study on developing
Buy an MAS standard for ensuring interoperability between agents
Pk,t , PSell
k,t Buying/selling power of MG k at t.
PRec Send of different designers has been carried out by [14] and [15].
t , Pt Receiving/sending power at t by MGs.
CDGP
Pt Output power of CDGP at t. Distributed MAS-based scheduling for islanded MGs has
min _adj
Pk,i,t Minimum adjustable power of unit i. been analyzed by [16] and [17]. Management of distributed
max _adj power sources using an MAS has been discussed in [18],
Pk,i,t Maximum adjustable power of unit i. distributed management of MGs by using MAS has been
proposed by [19] and [20] and MAS-based load shedding
strategy for MGs has been proposed by [21]. In order to
I. I NTRODUCTION overcome the drawbacks of centralized and distributed EMSs,
ROWING energy demand, environmental protection, various hybrid and hierarchical EMS systems have also been
G resiliency of the power grid in emergency situations, and
integration of distributed generations (DGs) along with energy
analyzed in the literature. MAS-based hybrid EMS systems
have been proposed by [22]–[24]. Apart from distributed
storage systems (ESSs) are the major considerations for the and centralized energy management strategies, currently var-
modern power industry. Microgrids (MGs) have the potential ious cooperative MMG systems have also gained popularity
to play a vital role in this organic transformation of the existing as in [25]–[27].
power grid to the future smart grid. MGs normally operate in Most of the researches available in the literature for hierar-
grid-connected mode to maximize the profit and have the abil- chical management of energy in MMG systems only provide
ity to operate in islanded mode to enhance the reliability during information about the surplus and shortage amounts of energy
emergency and/or scheduled maintenance periods. Optimal to the community EMS (CEMS) at each time interval. CEMS
operation and planning of the future smart distribution sys- only optimizes the surplus and shortage amount from indi-
tems is challenging due to uncertainties in generation amount vidual MGs. Each local EMS does not have any information
of DGs, electricity market prices, load demand, and penetra- about the production cost of other MGs of the MMG network.
tion of electrical vehicles (EVs). These uncertainties can be This approach results in unnecessary trading of electricity with
handled to a large extent by combining several MGs to form the utility grid and causes an increase in operational cost of
a multi-microgrids (MMG) system. Various researches have MMG system.
been conducted in the recent years regarding the configura- In [28], agent-based architecture has been developed for
tion and architectures of the future MMG systems. The major trading and managing power considering DR programs. The
considerations of an MMG system are minimization of opera- intelligent bidding strategy based on continues double auc-
tion cost, preservation of customer privacy, and enhancement tion (CDA) has been used to allow customers to participate
in grid reliability. in DR programs. A multiagent framework for implementing
Energy management system (EMS) is used to optimally demand side management (DSM) has been suggested by [29]
schedule the power resources and ESS to fulfil the load to solve the problem of power imbalance by paying more
demands. The architecture of an EMS could be centralized, attention to DR. An agent-based EMS has been developed
decentralized, or hybrid depending on the considerations stated for multiple MGs including DR and distributed storage (DS).
earlier. Genetic algorithms, robust optimization, stochastic The agent-based EMS with DR & DS has been implemented to
optimization, and mixed integer linear programming (MILP) reduce the system peak demand and minimize the cost of elec-
have been widely used in the literature for optimal schedul- tricity by [30]. In [31], a two-level architecture for multiple
ing of MMGs. Different demand response (DR) programs can MGs has been presented by using MAS. The market agents
also be incorporated in MMG EMS to reshape the load profiles have been designed to participate in real-time bidding. The
of MGs. bidding actions have been simulated by using naive auction
Energy management of MGs based on a double layer coor- algorithm.
dinated control in both grid-connected and islanded mode has The optimization strategy proposed in this paper is more
been proposed by [1]. Energy management of MMGs based related to the researches carried out by [19] and [28]–[31].
on a decentralized EMS has been suggested by [2] and a cen- Due to the merits of hierarchical EMSs and potential appli-
tralized EMS has been proposed by [3]. A robust energy cation of MAS in microgrid operation, all the studies includ-
management strategy for uncertainty aware MGs has been ing the proposed optimization strategy have used hierarchi-
analyzed by [4] while, a hierarchical EMS has been proposed cal EMSs for modeling and MAS for realizing the opti-
by [5] for preserving the customer privacy. An optimal energy mal operation of MGs. However, the model proposed in
storage strategy for grid-connected MGs has been proposed this study is based on the concept of cooperative MMG
by [6] and effect of energy storage system in operation of operation in contrast to the competitive models proposed
islanded MGs has been analyzed by [7]. A DR model based by [19] and [28]–[31]. The objectives of competitive and coop-
on real time price (RTP) has been developed by [8] and DR erative models are different, i.e., the objective of cooperative
has been used to mitigate the impacts of EVs on smart distri- MMG system is to reduce the operation cost of the entire net-
bution systems by [9] and [10]. Multiagent systems (MASs) work while in competitive environment objective of each MG
have been widely used for optimization of MMG systems. is to maximize its own profit. In addition, both the models find
The power engineering applications where MAS is required, their applications in different environments and have their own
have been analyzed along with the technical issues and pros and cons [12], [25].
BUI et al.: MULTIAGENT-BASED HIERARCHICAL ENERGY MANAGEMENT STRATEGY FOR MMGs 1325
for carrying out one round of optimization is summarized (level 2 agents). Each of the external and MG-EMS agents has
in Fig. 4. the liberty to decide about their participation in the global opti-
mization. In a similar way, each MG-EMS agent will inform its
E. Interaction Between Different Agents local resources about the market price signals along with cfp
The interaction between different developed agents in the messages. The local elements of MGs also can accept/reject
proposed operation strategy is shown in Fig. 5. Firstly, a mes- the proposals received from their MG-EMSs. Based on the
sage is sent by community EMS agent to market agent to proposals received from its local elements, each MG-EMS
inquire about the market price signals. The market agent sends agent will decide either to participate in global optimization
the day-ahead buying and selling prices for each hour of or not. All the MG-EMS agents will inform the CEMS agent
the day to CEMS agent. The market agent sends the same about their participation in global optimization. In the same
price signals to external agents. These market price signals way, external agents will also decide and inform the CEMS
are accompanied by call for proposal (CFP) messages and are agent. CEMS can accept or reject the proposals from external
destined to MG-EMSs and community level resource agents and MG-EMS agents.
1328 IEEE TRANSACTIONS ON SMART GRID, VOL. 9, NO. 2, MARCH 2018
CEMS will inform the level 2 agents (community level 0 ≤ SOCtB ≤ 1 ∀t ∈ T (9)
resource agents and MG-EMS agents) about the accep- B
SOCt−1 B
= (1 − δB ) · SOCt−1 ∀t ∈ T (10)
tance/rejection of their proposals. Respective MG-EMS agents
will inform their local agents about the decisions of Pt ,t ≤ IFt ,
Sh max
Pt,t ≤ OFtmax ∀t ∈ T
Sh
(11)
CEMS. Local and external agents will get either their com- t ∈T t ∈T
t =t t =t
mitment scheduling or will remain idle in accordance with
the decisions of CEMS and MG-EMSs. Finally, all the local OFtmax = PtL_con ∀t ∈ T (12)
L_adj L_ fix
agents and external agents will inform their respective upper Pt = Pt +PtL_con + PSh
t ,t − PSh
t,t ∀t ∈ T (13)
level agents about their commitment statuses. Communication
t ∈T
t ∈T
between all the agents is realized through agent commu- t =t t =t
III. MILP-BASED M ODEL FOR DAY-A HEAD S CHEDULING Equation (2) shows the operating bounds for ith CDG unit.
Mathematical models have been developed for each of The start-up indicator for CDG is represented by (3), which
the three steps explained in previous sections. In step 1, is max constraint. Apparently, this constraint leads to non-
local optimization is carried out by respective MG-EMSs. linear problem formulation but it can be solved by using
Each MG-EMS decides surplus/shortage and adjustable power “IloMax” operator available in CPLEX [33]. Power gener-
at this stage. Step-wise developed mathematical models are ated by photovoltaic cell, wind turbine, committed CDG units,
explained in the following sections. The proposed model is shortage power, and discharged amount should be balanced
formulated for a 24-h scheduling horizon with a time interval with load, surplus, and charging amounts at each interval and
of t, which could be any uniform interval of time. However, in is depicted by (4). BESS has been taken as a load during
the proposed day-ahead scheduling model, t has been assumed the charging mode and as a source during the discharg-
to be one hour. ing mode. Therefore, the cost of electricity is incorporated
during the charging/discharging mode. The capital cost of
BESS is an important issue in the planning phase opti-
A. Step 1: Local Optimization
mization, but in day-ahead scheduling capital cost is not
The proposed optimization model aims to minimize the total considered generally [3], [23]. Equations (5)-(10) show the
operational cost of each microgrid. Equation (1) is the objec- constraints for charging, discharging, and SOC of BESS. The
tive function and it contains CDG generation cost, startup cost charging/ discharging loss, self-discharge rate of BESS, and
for CDG, price of buying, price of selling, and penalty for BTB converter losses are included in the BESS modeling
shifting load from t to t’ interval respectively. as proposed by [35] and [36]. The SOC of BESS in the
B
first interval (SOCt−1,t=1 ) is the initial SOC of BESS: SOC of
min CiCDG · PCDG
i,t + yi,t · CiSU
last interval of previous day. The SOC of BESS is updated in
i∈I t∈T
Buy
each interval as given by (8). The amount of load, which can
+ PRt · PShort
t − PRSell
t · PSur
t + vt,t · PSh
t,t be shifted from interval t to t’, is constrained by (11). The max-
t∈T t,t ∈T imum amount of shiftable load is constrained by the amount
t=t
of controllable load and is modeled by (12). The adjusted load
(1)
after applying DR (load shifting) can be computed by using
Subject to equation (13). The penalty for shifting load from interval t to t’
can be modeled by using (14).
ui,t · Pmin
i ≤ PCDG ≤ ui,t · Pmax ∀i ∈ I, t ∈ T (2) After completing local optimization by each MG-EMS at
i,t i
yi,t = max ui,t − ui,t−1 , 0 , ui,t ∈ {0, 1} ∀i ∈ I, t ∈ T step 1, each MG calculates surplus, shortage, and adjustable
(3) power amount according to Algorithm 1. The calculated values
are communicated to the CEMS agent.
PPV
t + PWT
t + PCDG
i,t + PShort
t + PB−
t
i∈I
L_adj
= Pt + PSur
t + Pt
B+
∀t ∈ T (4) B. Step 2: Global Optimization
Cap 1 1 After receiving the information about surplus, shortage, and
0 ≤ PB+
t ≤ PB · 1 − SOCt−1
B
· · ∀t ∈ T
1 − LB+ ηBBTB adjustable power from each MG-EMG agent and proposals
(5) from CDGP and CBESS agents, CEMS performs global opti-
Cap mization in this step. The objective function for step 2 will be
0≤ PB−
t ≤ PB · SOCt−1
B
· 1 − LB− · ηBBTB ∀t ∈ T (6)
different depending on the operation mode of CBESS.
0≤ t , Pt
PB− B+
≤ PBTB
B /ηB
BTB
∀t ∈ T (7)
⎛ ⎞ 1) Subservient Mode Model: As described earlier, in sub-
1 1 servient mode the charging and discharging decisions of
1 · · P B−
SOCtB = SOCt−1
B
− Cap
· ⎝ 1 − LB− ηBBTB t ⎠ CBESS are made by CEMS and CBESS only follows the
PB −Pt · 1 − L
B+ B+
· ηB
BTB
received commands. The objective function (15) contains aver-
(8) age generation cost of CDGP, adjustable power cost from each
BUI et al.: MULTIAGENT-BASED HIERARCHICAL ENERGY MANAGEMENT STRATEGY FOR MMGs 1329
GP ≤ Pt
Pmin ≤ Pmax ∀t ∈ T
CDGP
GP (16)
min _adj Adj max _adj
Pk,i,t ≤ Pk,i,t ≤ Pk,i,t ∀k ∈ K, i ∈ I, t ∈ T C. Step 3: Local Optimization (Rescheduling)
(17) After performing global optimization, CEMS agent informs
Buy Adj
PCDGP
t + Pk,t + Pk,i,t + k,t + Pt
PSur CB− all the external and MG-EMS agents about their commitments
k∈K k∈K i∈I k∈K and scheduling. After receiving the global optimization results
= PShort
k,t + PCB+
t + PSell
k,t ∀t ∈ T (18) from CEMS agent, each MG-EMS performs local optimization
k∈K k∈K again (rescheduling). The objective function for reschedul-
0≤ CB+
Pt ≤
Cap
PCB · 1 − SOCt−1
CB ing is given by (24). Equation (24) contains CDG generation
amount with adjustable power, startup cost, prices for buying
1 1
× · BTB ∀t ∈ T (19) and selling power from/to the power grid, price for receiving
1−L CB+ ηCB from other MGs and price for sending power to other MGs of
Cap 1 the MMG system.
0 ≤ PCB−
t ≤ PCB · SOCt−1
CB
· (1 − LCB− ) · ∀t ∈ T
ηCB
BTB
(20) Adj
min CiCDG · PCDG
i,t + P i,t + yi,t · Ci
SU
0 ≤ PCB−
t , PCB+
t ≤ PBTB
CB /ηCB
BTB
⎛ ∀t ∈ T (21)
⎞ i∈I t∈T
1 1 Buy Buy
1 ⎜ · · PCB−
t ⎟ + PRt · Pt (t) − PRSell · PSell
· ⎝ 1 − L ηCB
CB− BTB t t
SOCtCB = SOCt−1
CB
− ⎠ t∈T
Cap
PCB −PtCB+
· 1−L CB+
· ηCB
BTB
+ PRRec
t · PRec
t (t) − PRt
Send
· PSend
t (26)
(22) t∈T
1330 IEEE TRANSACTIONS ON SMART GRID, VOL. 9, NO. 2, MARCH 2018
TABLE II
PARAMETER OF BESS AND CBESS IN THE MMG S YSTEM
Fig. 6. Buying and selling price along with generation cost for CDGs
TABLE III
of MGs.
S HORTAGE , S URPLUS A MOUNT IN E ACH MG OF MMG S YSTEM
Fig. 7. Load profiles of MG1 (a), MG2 (b) and MG3 (c) in response to DR.
TABLE I
PARAMETER OF CDG S IN E ACH MG OF MMG S YSTEM
Subject to
Adj
Buy
t + Pt
PPV + + Pi,t + Pt + PB−
t + Pt
WT
PCDG
i,t
Rec
i∈I
L_adj
= Pt + PSell
t + PB+
t + Pt
Send
∀t ∈ T (27) B. Step 1: Local Optimization
It can be observed from Fig. 6 that, the peak price interval
In addition to (27); load-balancing constraint, equation (26)
is from t=10 to t=21. Load from this interval will be shifted
is also constrained to (2)-(10). In load balancing, adjustable
to the off-peak intervals after applying DR in step 1. Load pro-
power and trading among MGs (sending and receiving) is also
files of MGs before and after the application of DR programs
considered as depicted by (25).
are illustrated by Fig. 7.
Fig. 7 shows that each of the MG is shifting its load from
IV. N UMERICAL S IMULATIONS peak time intervals to off-peak time intervals. Apart from the
A. Simulation Environment application of DR, shortage, surplus, and adjustable power
amounts are also determined after local optimization in step 1.
All MILP-based models developed for simulations of the MG-wise, shortage and surplus amounts for each MG are at
proposed operation of MMG system have been simulated in each interval of time is shown in tables III while upper and
Java, JADE with integration of IBM ILOG CPLEX. MCNP lower bounds for adjustable amounts are shown in table IV.
has been used with ACL for communication between various
agents developed for accomplishing the proposed algorithm.
The hour-wise buying and selling prices along with generation C. Step 2: Global Optimization
costs for CDGs of each MG are shown in Fig. 6. Generation CEMS receives the information shown in tables III and IV
and start-up costs along with generation limits of CDG units from each of the MG and average per-unit cost from
in each MG of the MMG system are shown in Table I. In CDGP. Depending on the mode of CBESS, charging and
the proposed MMG optimization strategy, BESSs in MGs and discharging of CBESS is decided. The test case results for
CBESS are connected to the MMG system through back- charging/discharging of CBESS in subservient mode (SM) and
to-back (BTB) converters. Parameters related to CBESS and autonomous mode (AM) are shown in table V. The charg-
BESSs of individual MGs are tabulated in Table II. ing/discharging of CBESS in SM is decided by CEMS for
BUI et al.: MULTIAGENT-BASED HIERARCHICAL ENERGY MANAGEMENT STRATEGY FOR MMGs 1331
TABLE IV TABLE V
B OUNDS FOR A DJUSTABLE A MOUNT IN E ACH MG OF MMG S YSTEM CBESS C HARGING AND D ISCHARGING S CHEDULE IN SM AND AM
Fig. 8. Buying and selling amount of electricity in case 1 and case 2. Fig. 10. Difference of buying, selling, and adjustable power in CBESS
modes.
surplus and shortage amounts. The mismatch of total surplus [13] S. D. J. McArthur et al., “Multi-agent systems for power engineering
and shortage power has been not only compensated by trading applications—Part I: Concepts, approaches, and technical challenges,”
IEEE Trans. Power Syst., vol. 22, no. 4, pp. 1743–1752, Nov. 2007.
power with utility grid or external resource (CBESS/CDGP) [14] S. D. J. McArthur et al., “Multi-agent systems for power engineering
but also can be adjusted by exchanging power with other applications—Part II: Technologies, standards, and tools for build-
MGs. Being a cooperative model, the trading among MGs ing multi-agent systems,” IEEE Trans. Power Syst., vol. 22, no. 4,
pp. 1753–1759, Nov. 2007.
has been increased by increasing the power of cheaper CDGs [15] L. Wang, Z. Wang, and R. Yang, “Intelligent multiagent control system
having adjustable power. This trading has resulted in reduc- for energy and comfort management in smart and sustainable buildings,”
tion of external trading and operation cost of MMG system. IEEE Trans. Smart Grid, vol. 3, no. 2, pp. 605–617, Jun. 2012.
[16] T. Logenthiran, D. Srinivasan, and A. M. Khambadkone, “Multi-agent
The proposed strategy has proved to be more cost efficient system for energy resource scheduling of integrated microgrids in a dis-
as compared to the prevailing energy management strategies. tributed system,” Elect. Power Syst. Res., vol. 81, no. 1, pp. 138–148,
All the developed models are based on MILP, which makes it Jan. 2011.
[17] H.-M. Kim, T. Kinoshita, and M.-C. Shin, “A multiagent system for
easy to implement and computationally inexpensive. DR has autonomous operation of islanded microgrids based on a power market
also been applied to MMG network and the effects of DR environment,” Energies, vol. 3, no. 12, pp. 1972–1990, Dec. 2010.
are evident from the simulation results. Due to the hierarchi- [18] J. Lagorse, D. Paire, and A. Miraoui, “A multi-agent system for energy
management of distributed power sources,” Renew. Energy, vol. 35,
cal EMS, the proposed optimization strategy distributes the no. 1, pp. 174–182, Jan. 2010.
computational burden and yet makes the optimal energy man- [19] Y. S. F. Eddy, H. B. Gooi, and S. X. Chen, “Multi-agent system for dis-
agement of MMGs possible. Simulation studies have shown tributed management of microgrids,” IEEE Trans. Power Syst., vol. 30,
no. 1, pp. 24–34, Jan. 2015.
that, in order to reduce the overall operating cost of MMG [20] H.-M. Kim, Y. Lim, and T. Kinoshita, “An intelligent multiagent sys-
system, the subservient mode of CBESS is more effective. tem for autonomous microgrid operation,” Energies, vol. 5, no. 9,
CBESS can increase its benefit in autonomous operational pp. 3347–3362, Sep. 2012.
[21] Y. Lim, H.-M. Kim, and T. Kinoshita, “Distributed load-shedding sys-
mode at the cost of elevated operational cost of the entire tem for agent-based autonomous microgrid operations,” Energies, vol. 7,
MMG system. no. 1, pp. 385–401, 2014.
[22] M. Mao, P. Jin, N. D. Hatziargyriou, and L. Chang, “Multiagent-based
hybrid energy management system for microgrids,” IEEE Trans. Sustain.
Energy, vol. 5, no. 3, pp. 938–946, Jul. 2014.
R EFERENCES [23] C.-X. Dou and B. Liu, “Multi-agent based hierarchical hybrid control for
smart microgrid,” IEEE Trans. Smart Grid, vol. 4, no. 2, pp. 771–778,
[1] Q. Jiang, M. Xue, and G. Geng, “Energy management of microgrid Jun. 2013.
in grid-connected and stand-alone modes,” IEEE Trans. Power Syst., [24] C. M. Colson, M. H. Nehrir, R. K. Sharma, and B. Asghari, “Improving
vol. 28, no. 3, pp. 3380–3389, Aug. 2013. sustainability of hybrid energy systems part II: Managing multiple objec-
[2] Z. Wang, B. Chen, J. Wang, and J. Kim, “Decentralized energy tives with a multiagent system,” IEEE Trans. Sustain. Energy, vol. 5,
management system for networked microgrids in grid-connected and no. 1, pp. 46–54, Jan. 2014.
islanded modes,” IEEE Trans. Smart Grid, vol. 7, no. 2, pp. 1097–1105, [25] T. Morstyn, B. Hredzak, and V. G. Agelidis, “Cooperative multi-agent
Mar. 2016. control of heterogeneous storage devices distributed in a DC microgrid,”
[3] D. E. Olivares, C. A. Cañizares, and M. Kazerani, “A centralized energy IEEE Trans. Power Syst., vol. 31, no. 4, pp. 2974–2986, Jul. 2016.
management system for isolated microgrids,” IEEE Trans. Smart Grid, [26] N.-O. Song, J.-H. Lee, H.-M. Kim, Y.-H. Im, and J.-Y. Lee, “Optimal
vol. 5, no. 4, pp. 1864–1875, Jul. 2014. energy management of multi-microgrids with sequentially coordinated
[4] Y. Zhang, N. Gatsis, and G. B. Giannakis, “Robust energy management operations,” Energies, vol. 8, no. 8, pp. 8371–8390, Aug. 2015.
for microgrids with high-penetration renewables,” IEEE Trans. Sustain. [27] E. M. Hammad, A. K. Farraj, and D. Kundur, “Grid-independent coop-
Energy, vol. 4, no. 4, pp. 944–953, Oct. 2013. erative microgrid networks with high renewable penetration,” in Proc.
[5] Z. Wang, K. Yang, and X. Wang, “Privacy-preserving energy schedul- IEEE Power Energy Soc. Innov. Smart Grid Technol. Conf. (ISGT),
ing in microgrid systems,” IEEE Trans. Smart Grid, vol. 4, no. 4, Washington, DC, USA, Feb. 2015, pp. 1–5.
pp. 1810–1820, Dec. 2013. [28] H. S. V. S. K. Nunna and S. Doolla, “Demand response in smart dis-
[6] P. Malysz, S. Sirouspour, and A. Emadi, “An optimal energy storage tribution system with multiple microgrids,” IEEE Trans. Smart Grid,
control strategy for grid-connected microgrids,” IEEE Trans. Smart Grid, vol. 3, no. 4, pp. 1641–1649, Dec. 2012.
vol. 5, no. 4, pp. 1785–1796, Jul. 2014. [29] H. S. V. S. K. Nunna and S. Doolla, “Responsive end-user-based demand
side management in multimicrogrid environment,” IEEE Trans. Ind.
[7] H. HassanzadehFard, S. M. Moghaddas-Tafreshi, and S. M. Hakimi,
Informat., vol. 10, no. 2, pp. 1262–1272, May 2014.
“Effect of energy storage systems on optimal sizing of islanded micro-
[30] H. S. V. S. K. Nunna and S. Doolla, “Energy management in micro-
grid considering interruptible loads,” in Proc. 3rd IEEE Int. Youth Conf.
grids using demand response and distributed storage—A multiagent
Energ. (IYCE), Leiria, Portugal, Jul. 2011, pp. 1–7.
approach,” IEEE Trans. Power Del., vol. 28, no. 2, pp. 939–947,
[8] A. J. Conejo, J. M. Morales, and L. Baringo, “Real-time demand
Apr. 2013.
response model,” IEEE Trans. Smart Grid, vol. 1, no. 3, pp. 236–242,
[31] H. S. V. S. K. Nunna and S. Doolla, “Multiagent-based distributed-
Dec. 2010.
energy-resource management for intelligent microgrids,” IEEE Trans.
[9] S. Shao, M. Pipattanasomporn, and S. Rahman, “Demand response as Ind. Electron., vol. 60, no. 4, pp. 1678–1687, Apr. 2013.
a load shaping tool in an intelligent grid with electric vehicles,” IEEE [32] Java Agent Development Environment (JADE). Accessed on Jul. 2016.
Trans. Smart Grid, vol. 2, no. 4, pp. 624–631, Dec. 2011. [Online]. Available: http://jade.tilab.com
[10] Z. Zhou, F. Zhao, and J. Wang, “Agent-based electricity market simu- [33] IBM ILOG CPLEX V12.1 User’s Manual for CPLEX 2009, CPLEX
lation with demand response from commercial buildings,” IEEE Trans. Division, ILOG, Incline Village, NV, USA, 2009.
Smart Grid, vol. 2, no. 4, pp. 580–588, Dec. 2011. [34] FIPA. The Foundation for Intelligent Physical Agents Standards.
[11] T. Logenthiran, D. Srinivasan, A. M. Khambadkone, and H. N. Aung, Accessed on Jul. 2016. [Online]. Available: http://www.fipa.org
“Multiagent system for real-time operation of a microgrid in real-time [35] B. Zhao, X. Zhang, J. Chen, C. Wang, and L. Guo, “Operation opti-
digital simulator,” IEEE Trans. Smart Grid, vol. 3, no. 2, pp. 925–933, mization of standalone microgrids considering lifetime characteristics
Jun. 2012. of battery energy storage system,” IEEE Trans. Sustain. Energy, vol. 4,
[12] M. H. Cintuglu, H. Martin, and O. A. Mohammed, “Real-time imple- no. 4, pp. 934–943, Oct. 2013.
mentation of multiagent-based game theory reverse auction model for [36] M. Choobineh, P. C. Tabares-Velasco, and S. Mohagheghi, “Optimal
microgrid market operation,” IEEE Trans. Smart Grid, vol. 6, no. 2, energy management of a distribution network during the course of a heat
pp. 1064–1072, Mar. 2015. wave,” Elect. Power Syst. Res., vol. 130, pp. 230–240, Jan. 2016.
BUI et al.: MULTIAGENT-BASED HIERARCHICAL ENERGY MANAGEMENT STRATEGY FOR MMGs 1333
Van-Hai Bui received the B.S. degree in electrical Hak-Man Kim received the first Ph.D. degree
engineering from the Hanoi University of Science in electrical engineering from Sungkyunkwan
and Technology, Vietnam, in 2013. He is currently University, South Korea, in 1998, and the second
pursuing the combined master’s and Ph.D. degrees Ph.D. degree in information sciences from Tohoku
with the Department of Electrical Engineering, University, Japan, in 2011. He was with Korea
Incheon National University, South Korea. His Electrotechnology Research Institute, South Korea,
research interests include microgrid operation and from 1996 to 2008. He is currently a Professor with
energy management system. the Department of Electrical Engineering, Incheon
National University, South Korea. His research inter-
ests include microgrid operation and control and dc
power systems.
Akhtar Hussain received the B.E. degree in
telecommunications from the National University
of Sciences and Technology, Pakistan, in 2011,
and the M.S. degree in electrical engineering from
Myongji University, Yongin, South Korea, in 2014.
He is currently pursuing the Ph.D. degree with
Incheon National University, South Korea. He was
an Associate Engineer in SANION, South Korea,
from 2014 to 2015. His research interests are distri-
bution automation and protection, smart grids, and
microgrid optimization.