2019 International Conference on Intelligent Green Building and Smart Grid (IGBSG2019), 6-9 Sept.

, Yichang
China

A Day-ahead Scheduling Optimization Model of

Multi-Microgrid Considering Interactive Power
Control
Lijun He, Zhaobin Wei,Hai Yan, Kang-Yi Xv, Meng-yu Zhao, Shan Cheng
Hubei Collaborative Innovation Center for Microgrid of New Energy
China Three Gorges University
Yichang, China
[email protected]

Abstract—Aiming at the problem of insufficient stability dispatching mechanism underlying optimization problem
and security considerations for multi-microgrid system is coordinated, but the security and reliability of the
access distribution network, this paper proposes a multi- system operation are not considered, and no specific
microgrid day-ahead scheduling optimization model trading mechanism is designed. The work in [9] uses
considering interactive power control. The upper cooperative game theory to solve the problem of bidding
optimization model aims to reduce the interaction power and income distribution among MGs and proposes method
between the microgrid and the distribution network and for judging identity duality. The work in [10] proposed a
coordinates the output between the microgrids. The lower market trading model based on cooperative game theory
optimization model is devoted to solve the economic
for the photovoltaic MG group, but did not mention the
optimization operation of the multi-microgrid system. The
bilateral bidding strategy is adopted to make the microgrid
specific scheduling method.
conduct energy trading to reduce the overall cost of the In view of the shortcomings of existing literatures on the
multi-microgrid system and reduce the impact of power optimization scheduling method of MMS, this paper
fluctuations on the distribution network. The distribution proposes a two-layer optimization model considering
network system with multiple microgrids is used to verify interactive power control, and reduces the interaction
that the optimization model can obtain better output plans power and interaction power fluctuation between MGs
for each microgrid. The set trading strategy encourages and distribution network through the upper distribution
effectively the mutual energy between the microgrids and network optimization model. The underlying microgrid
reduces the micro-grid’s impact of grid operation on the optimization model is based on this result to coordinate
distribution network. the output between the microgrids through the designed
bilateral bidding trading strategy, so that this model can
Keywords—Multi-Microgrid, Day-ahead Optimization encourage active power trading between the MGs instead
Scheduling, Bilateral Bidding, Power Market Introduction of directly dealing with the main grid. The operation of
the MMS in the distribution network is simulated to verify
I. INTRODUCTION the effectiveness of the scheduling method.
In recent years, large-scale distributed power sources II. TRADING FRAMEWORK BASED ON TWO-LAYER
have been widely integrated into power systems in the OPTIMIZATION
form of microgrids(MGs). At the same time, with the The MMS concerned in this paper refers to multiple
promotion of MG technology, more and more MGs access interconnected MGs in the region accessing the same
regional distribution systems form local multi-microgrid medium-low voltage distribution system. MGs can deal
system(MMS) [1-4]. However, the integration of MMS with each other through the seller and buyer aggregator,
complicates the interests and makes it difficult to and they unify into the power dispatch center for
guarantee the operational safety of the system, which scheduling ultimately.
poses new challenges for power system optimization
scheduling [5]. In order to ensure the stability, reliability Dispatch center Distribution network
and economic benefits of MMS access to power systems,
it is necessary to further explore the stable operation
strategy of MMS and new market trading mechanisms.
There are many studies on the scheduling of MMS. In Buyer aggregator Seller aggregator
[6], a MMS network coordinated control strategy based on
predictive control is proposed. The multi-objective
optimization model is constructed considering the network
loss and load fluctuation variance, but the power
interaction between MGs is not considered. Reference [7]
… …
proposes a multi-objective optimization scheduling MG1 MGm MGm+1 MGn
method for regional MMS network with economy and
comprehensive energy efficiency. The work in [8] Buyer microgrids Seller microgrids
proposed a two-layer scheduling method for MMS in open
market environment. The lower layer model optimizes the Energy Flow
benefits of each MG and the upper-level distribution Information Flow
This research was funded by the Hubei Provincial Department of
Education Science and Technology Research Project(Q20161203).

XXX-X-XXXX-XXXX-X/XX/$XX.00 ©20XX IEEE

666
 2019 International Conference on Intelligent Green Building and Smart Grid (IGBSG2019), 6-9 Sept., Yichang
China

Fig.1 Multi-microgrid system transaction structure 24 m

Wi ,environment   Ck  k Pi ,MT
t (5)
The complex system formed by the same power t 1 k 1
distribution dispatching mechanism is shown in Fig.1. where Ck is the unit cost required to treat the kth pollutant.
Each MG includes wind power (WT), photovoltaic (PV), δk represents the kth pollutant emission factor of the power
micro gas turbine (MT), battery(BT), and load. generation unit.
The MMS proposed in this paper will conduct
III. OPTIMIZATION MODEL
transactions between the MGs, so that the transaction
A. Upper Optimization Model costs are calculated as follows:
24
Due to the uncertainty of the renewable generation, the Wi ,deal   (Cgird
sell
Pt ,grid,i  Cgrid
buy
Pt ,grid,i
internal load of the MMS may lack power supply or t 1 (6)
excess power, which will generate interactive power with
the distribution network. In order to ensure the stable C deal deal
i ,tP i ,t  aservice  bservice )
operation of the distribution network when the MMS is where Csell
grid is the price which MG i purchases electricity
connected to the distribution network, the interaction from the distribution grid, Cbuygrid is the price at which MG i
power and interaction power fluctuations of the above- sells electricity to the distribution grid, Cdeali is the price
mentioned distribution network and MMS are optimized which MG i sells/purchases to other MGs. aservice represents
for the upper layer. the service fee payable to the aggregator between the
In order to reduce the interaction power between the MGs’ transactions, bservice represents the service fee payable
MGs and the distribution network, the following for the transaction between the MG and the distribution
optimization objectives are set(for ∀ t,i∈I): network.
24 n
f1 = Pi ,Lt  Pi ,WT
t  Pi ,MT
t  Pi ,BT
t  Pi ,t
PV
(1) C.Restrictions
t 1 i 1
1) Power balance constraints:
where PLi,t is the total load of the MG i during time t, PPV
i,t , P
WT MT BT
i,t ,Pi,t ,Pi,t are respectively the power supplied by the MG
Pi ,Lt  Pi ,PV
t  Pi ,t
WT
 Pi ,BT
t  Pt ,grid,i (7)
i during the time t. 2) Interactive power constraints:
For reducing the total power exchange fluctuation level
between the distribution network and the MG, the MG is
min
Pgrid,i
 Pt ,grid,i  Pgrid,
max
i (8)
involved in the regulation of load peak-to-valley 3) Distributed power output constraints:
adjustment of the distribution network, and the power
fluctuation is taken as the objective function: Pi ,min
t
 Pi .t  Pi ,max
t (9)
24 n
4) Battery capacity constraints:
 ( P t ,gird,i  Pt 1,gird,i )2
Eimin  Ei ,t  Eimax
f2  t  2 i 1
(2) ,t ,t (10)
T
Ei ,t 1  Ei ,t  P t BT
where Pt,grid,i is the interaction power of the MG i with the i ,t
(11)
distribution network during the time t, and T is the total 5) MT climbing rate constraints:
number of time periods, where T=24. The multi-objective  RMT,d t  Pi ,MT
t 1  Pi ,t  RMT,u t
MT
(12)
optimization model is normalized and weighted by the
entropy weight method. The final optimization target is IV. BILATERAL BIDDING STRATEGY BETWEEN MGS
transformed into:
F  1  F1  2  F2 (3) A. Transaction Matching Rule
B. Lower Optimization Model In order to encourage active energy trading between
MGs, this paper designs the following bilateral bidding
In order to achieve the economic operation of the MMS, rules for energy trading to reduce the energy interaction
the lower-level model optimizes the MMS based on the with the distribution network and promote the mutual
optimized distribution power of the upper-level energy exchange between the MGs:
distribution network. It aims to optimize the overall 1) Judging by the formula (13) whether the MG lacks
efficiency of each MG in the MMS, mainly including
electricity:
operation management costs, MT generation costs,
environmental costs, and MG transaction costs. Pi ,t  Pi ,Lt  Pi ,Gt (13)
Taking into account the operation and management costs G
Where P i,t represents the output power that the MG i
of power generation units in each MG, the O&M cost is itself participates in to supply the load; Pdeal i,t represents the
calculated by the following formula including the electricity purchased or sold by the MG i; if Pi,t<0, the MG
operating and management costs of PV, WT, MT, and BT: i is deficient MG; if Pi,t>0, the MG i is a surplus MG;
24
2) Transactions:
Wi ,OM   ( K OM,PV Pi ,PV
t  K OM,WT Pi ,t
WT
During the time interval, the buyer MG reports the
t 1 (4)
power shortage and the respective expected purchase price
K MT
OM,MT i ,t P K BT
OM,BT i ,t P ) c to the buyer aggregator. The seller MG will report the
where: K OM,PV , K OM,WT , K OM,MT , K OM,BT are the operating redundant power and the respective expected electricity
management coefficients of PV, WT, MT, BT. price d to the seller aggregator. The "high and low
The main sources of pollutants discharged from MMS is matching" trading rules are adopted, that is the buyer
MT, and its environmental cost is calculated as follows: aggregator sorts the price according to the price of the

667
 2019 International Conference on Intelligent Green Building and Smart Grid (IGBSG2019), 6-9 Sept., Yichang
China

buyer MG from high to low and the seller aggregator sorts swarm optimization algorithm to solve the optimization
the price according to the price of the seller MG from low problem.
to high. This rule combines the power generation required There are three strategies set for comparative analysis.
by the highest priority buyer and seller. The transaction Strategy 1: The MG is limited to interacting with the
price is the average of the expected price of both bids after distribution network, regardless of the transactions
the match is successful, expressed as between the MGs;
c  d  Strategy 2: Adopt the trading strategy, but do not
Cideal
,t  (14) perform double-layer optimization;
2 Strategy 3: Adopt the trading strategy and perform
This paper considers the pricing of new energy subsidies, double-layer optimization.
natural gas prices and energy storage costs, and the
buyer's quotation is not higher than the purchase price of
electric power to the distribution network. The seller's
quotation is not lower than the electricity price to the
distribution network.
B. Double Layer Optimization With Bilateral Bidding
Trading Strategy
In MMS, the power-deficient MG is registered as a
buyer in the buyer’s aggregator and reports the lack of
electricity. The surplus MG is registered as a seller in the
seller aggregator and reports excess power. The buyer
aggregator passes the reasonable trading strategy with the
seller. The aggregator exchanges information and
concludes the transaction, and then submits the transaction
volume to the power dispatching center. The power
dispatching center uses the WT, PV and load forecast data
Fig.3 Three strategies of the interaction power between the microgrid
submitted by each MG to form a dispatch plan. and the distribution network
Initialization
By comparing the total interaction power of the three
MGs and the distribution network in the three strategies,
WT and PV forecasting, load forecasting the total interaction power of Strategy 3 is obviously
smaller than the strategy one or two during the peak
period. Moreover, the transmission to the large grid is
The upper dispatch center gets the data and optimizes
reduced by the BT charging during the low valley period.
The better peak-shaving effect indicates that the double-
The lower seller and the buyer are registered at the corresponding layer optimization model set in this paper can effectively
aggregator and comply with the trading strategy
reduce the interactive power and interactive power
fluctuations between the MG and the distribution network,
Lower level optimization model objective function calculation and improve the reliability of the MG operation.

Whether to meet the constraints of each MG

Form a scheduling plan

Whether to meet the interactive power constraints

Form the final

scheduling plan

Fig.2 Double-layer flow chart with trading strategy

V. CASE STUDIES
In this paper, a distribution network with three MGs is
selected for 24 time intervals. Each MG distributed power
supply parameters and load prediction are in reference [7],
and the time-of-use electricity price is adopted for
distribution network electricity/purchasing electricity Fig.4 Trading between microgrids
price. The MG is not allowed to buy and sell at the same
time. The weighting factor ϕ1=ϕ2=0.5 in the upper-level
optimization model. This paper uses the improved particle

668
 2019 International Conference on Intelligent Green Building and Smart Grid (IGBSG2019), 6-9 Sept., Yichang
China

Fig.4 shows the transactions between the MGs in game trading method can promote the power transaction
Strategy 3. It can be seen from the figure that the MG 2 between MGs more effectively.
lacks electricity mainly because of the purchase of
electricity, and the MG 1 and the MG 3 have more excess REFERENCES
electricity, mainly for electricity sales transactions, and [1] C.S. Wang, Z. Wu, P. Li. Research on Key Technologies of
are sold to the MG 2 . It can be seen that the seller is more Microgrid[J]. Journal of Electric Engineering, 2014, 29(2): 1-12.
inclined to conclude a deal with the buyer who can [2] Q.S. Chen, X.J. Wang, W. Chi, et al. Research on Energy
purchase the most electricity, and the trading strategy can Management Method of Multi-microgrid Interconnected System[J].
Power System Protection and Control, 2018, 46(11): 83-91.
promote the energy consumption of the new energy
[3] T. Lu, Q. Ai, Z.Wang. Interactive game vector:A stochastic
generation and the energy transaction between the MGs. operation-based pricing mechanism for smart energy systems with
coupled-microgrids[J]. Applied Energy, 2018, 212:1462-1475.
[4] Y. Bai, A. Li, Q. Xia. Electricity business marketing modes in the
new environment and new electricity pricing systems[J]. Power
System Protection and Control, 2016, 44(5): 10-16.
[5] Y.Z. Zhou, H. Wu, Y.I. Li, H.H. Xin, Y.H. Song. Dynamic
scheduling of neighboring island multi-microgrid network based on
MCS-PSO algorithm[J]. Automation of Electric Power Systems,
2014, 38(09): 204-210.
[6] X. Shen Xin, M. Cao, R. Yan, et al. Research on coordinated
optimization scheduling and economic operation of multi-micro
distribution system[J]. Journal of Electronic Measurement and
Instrument, 2016, 30(4): 568-576.
[7] S.X. Wang, Z.J. Wu, S.C. Yuan, et al. Multi-objective optimization
scheduling method for regional multi-microgrid systems[J].
Electric Power System and Automation, 2017, 29(5): 14-20.
[8] W. Wei, H.N. Sun, F.Z. Luo, et al. Optimal scheduling method for
multi-microgrid systems based on open market environment[J].
Electric Power System and Automation, 2016, 28(7): 19-25.
Fig.5 Total benefits per microgrid per period [9] P.L. Bai, X. Lei, J.P. He. Design of BiddingMechanism for
Multiple Microgrids[J]. Modern Electric Power. 2017(6):22-27.
As shown in Fig.5 and Tab.1, the total cost of strategy [10] N. Liu, W. Zhao, J. Wang, et al. Photovoltaic microgrid group
two and three is far less than that of strategy one. This is trading model based on cooperative game theory[J]. Transactions
because the MG conducts bidding transactions, which of China Electrotechnical Society, 2018, 33(8): 1903-1910.
promotes the mutual energy exchange between the MGs,
reduces the pressure on the MG 2, 3 to supply more
electricity to the distribution network, and brings more to
the MG 1. The gains are win-win. It is proved that the
bidding strategy can promote the reasonable distribution
of electric energy in the MMS and ensure that each MG
can obtain operating income beyond its independent
operation.

TABLE Ⅰ. MGS TOTAL COST

Total cost Strategy1 Strategy2 Strategy3

MG1 -236263 -224559 -220909

MG2 314227 300909 300160

MG3 185742 164048 166611

Total 263706 240398 245862

VI. CONCLUSIONS
In this paper, a day-ahead scheduling of MMS
optimization model considering interactive power control
is proposed. The following results can be obtained from
the operation optimization results of distribution network
systems with MMS. The two-layer optimization model
proposed in this paper can help to improve the optimal
scheduling of MMS and improve the stability and security
of the system operation. Adopting the trading strategy set
in this paper can improve the economics of MG operation
and improve the new energy consumption level of MG.
Furthermore, the weight factor in the upper-level model
should be discussed, and for MG or distributed power
generation with different stakeholders, adopting a better

669