Power distribution network-shared energy storage-multi-microgrid optimization scheduling method based on multi-agent game

Abstract

The invention discloses a power distribution network-shared energy storage-multi-microgrid optimization scheduling method based on a multi-agent game. The method specifically comprises the following steps: constructing a multi-microgrid shared energy storage multi-target optimization configuration model considering energy storage investment cost and load fluctuation; solving the multi-microgrid shared energy storage multi-target optimization configuration model by adopting a multi-target particle swarm algorithm, and optimizing to obtain charging and discharging power of energy storage to each microgrid; establishing a one-master and two-slave game scheduling model taking the power distribution network as a master and multiple microgrids and shared energy storage as followers; and solving the one-master two-slave game scheduling model by adopting a genetic algorithm nested solver to obtain an optimal scheduling scheme. By constructing a one-master two-slave multi-agent game scheduling framework, cooperation and coordination among the power distribution network, the shared energy storage and the multiple microgrids are facilitated, coordination of three-party scheduling is realized, balanced distribution of benefits of the power distribution network, the shared energy storage and the multiple microgrids is further realized, and the overall benefit is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of power system optimization and dispatching, and specifically relates to a distribution network-shared energy storage-multi-microgrid optimization and dispatching method based on multi-agent game. Background Art

[0002] Under the background of the "dual carbon" strategy, integrated energy microgrids have achieved rapid development with their significant advantages in clean energy substitution and promoting low-carbon sustainable development of energy. The optimization and dispatching of integrated energy microgrids has also become a research focus. With the continuous increase in the number of system microgrids, the new development pattern of multi-microgrid joint dispatch has gradually emerged. However, the coupling and interaction mechanism between multi-microgrids and multi-energy flows is complex, and the superposition of multiple uncertainties such as source and load has brought unprecedented challenges to traditional dispatching methods.

[0003] At present, domestic and foreign scholars have proposed many new scheduling methods, but their scheduling of shared energy storage participating in multiple microgrids focuses on the capacity configuration and operation strategy of shared energy storage, fails to consider the issue of interest distribution between shared energy storage and multiple microgrids, and does not seek a balance between using energy storage to smooth fluctuations and improving economic efficiency.

[0004] When solving energy trading problems, game theory, as an effective tool, provides valuable ideas for the difficult problem of interest distribution among multiple subjects. However, when discussing the game relationship between multi-microgrids and shared energy storage, the existing methods fail to fully consider the participation of the distribution network, and fail to fully utilize the characteristics of multi-microgrid grid connection to include shared energy storage, multi-microgrids and distribution networks in the consideration of game scheduling. Summary of the invention

[0005] The purpose of the present invention is to provide a distribution network-shared energy storage-multi-microgrid optimization scheduling method based on multi-agent game, so as to achieve a balanced distribution of the interests of the distribution network, shared energy storage and multi-microgrids, thereby improving the overall operating efficiency.

[0006] The technical solution adopted by the present invention is a distribution network-shared energy storage-multi-microgrid optimization scheduling method based on multi-agent game, which is specifically implemented according to the following steps:

[0007] Step 1: Construct a multi-objective optimization configuration model for multi-microgrid shared energy storage that takes into account energy storage investment costs and load fluctuations;

[0008] Step 2, using a multi-objective particle swarm algorithm to solve the multi-objective optimization configuration model of multi-microgrid shared energy storage in step 1, and optimize the charging and discharging power of energy storage to each microgrid;

[0009] Step 3: Establish a one-master-two-slave game dispatching model with the distribution network as the leader and multiple microgrids and shared energy storage as followers;

[0010] Step 4: Based on the optimization of the charging and discharging power of the energy storage to each microgrid, the genetic algorithm nested solver is used to solve the one-master-two-slave game scheduling model to obtain the optimal scheduling solution.

[0011] The present invention is also characterized in that:

[0012] In step 1, specifically: step 1.1, establishing two optimization objectives of minimizing energy storage investment cost and minimizing net load standard deviation;

[0013] (1) The energy storage cost of microgrid i is the lowest, as shown in formula (1):

[0014] C SES,i =min(C INV,i +C OM,i ) (1);

[0015] Where: C SES.i is the energy storage investment cost of microgrid i, C INV.i is the energy storage investment cost borne by microgrid i, C OM.i Energy storage operation and maintenance costs borne by microgrid i;

[0016] 1) The energy storage investment cost borne by microgrid i is shown in formula (2):

[0017]

[0018] Where: P , E They represent the unit power investment cost and unit capacity investment cost of shared energy storage respectively; They represent the maximum values ​​of energy storage power and capacity required by microgrid i; T E represents the service life of the shared energy storage;

[0019] 2) The energy storage operation and maintenance cost borne by microgrid i is shown in formula (3):

[0020]

[0021] Where: P SES,C,i (t), P SES,D,i (t) respectively represent the charging and discharging power of energy storage to microgrid i in time period t; δ OM The unit power operation and maintenance cost of energy storage for microgrid i; T is the dispatch period; Δt is the time interval;

[0022] (2) The net load standard deviation of microgrid i is the smallest, as shown in equations (4)-(5):

[0023]

[0024] Step 1.2, determine the constraints of the two optimization objectives, and optimize the shared energy storage for the charging and discharging power of each microgrid; the constraints include the coupling constraints and the charging and discharging power constraints shared with microgrid i;

[0025] (1) The coupling constraint shared with microgrid i is shown in formula (6):

[0026]

[0027] is the upper limit of the energy storage power demand of microgrid i, as shown in formula (7):

[0028]

[0029] Where: They represent the upper limits of charging and discharging power of energy storage to microgrid i; U SES,C,i (t) is 1 for charging and 0 for discharging; U SES,D,i (t) 1 indicates discharging, and 0 indicates charging;

[0030] (2) Charge and discharge power constraints, as shown in equations (8)-(10):

[0031] E i (t) = E i (t-1)+(η SES,C P SES,C,i (t)-P SES,D,i (t) / η SES,D )Δt (8);

[0032]

[0033] In step 3, specifically:

[0034] (1) Establishing the main distribution network dispatching model

[0035] 1) Objective function

[0036] The decision variable for optimal dispatching of distribution network is the purchase and sale price of electricity, and the goal of optimal operation is the total cost C DN Minimum, as shown in formula (11):

[0037] C DN =min(C grid +C MMG,DN +C SES,DN ) (11);

[0038] (a) The cost of purchasing electricity from the main grid is shown in formula (12):

[0039]

[0040] Where: sell,grid (t) are the electricity prices of the main grid in time period t; P buy,DN (t) is the power purchased by the distribution network from the main grid during time period t;

[0041] (b) The cost of electricity interaction with multiple microgrids, as shown in formula (13):

[0042]

[0043] (c) The cost of electricity interaction with shared energy storage, as shown in formula (14):

[0044]

[0045] Where: P buy,SES (t), P sell,SES (t) is the power purchased and sold by the shared energy storage from the distribution network during period t;

[0046] 2) Constraints

[0047] (a) Power flow balance constraint, as shown in formula (15):

[0048]

[0049] (b) State variable constraints, as shown in formula (16):

[0050]

[0051] (c) Distributed energy access constraints, as shown in formula (17):

[0052]

[0053] (d) Time-of-use electricity price constraints;

[0054] The electricity purchase and sale prices set by the distribution network for microgrids and shared energy storage are time-of-use electricity prices. The electricity sales prices and electricity purchase prices during off-peak hours, normal hours, and peak hours satisfy the following relationship, as shown in equations (18)-(19):

[0055]

[0056] The electricity selling price of the distribution network in each period must be higher than the electricity purchasing price, as shown in formula (20):

[0057]

[0058] The average electricity price within the dispatch period T should not be greater than the upper limit of the distribution network electricity price, as shown in formula (21):

[0059]

[0060] (2) Establish a multi-microgrid dispatching model with shared energy storage;

[0061] 1) Establish a shared energy storage dispatch model

[0062] (a) Objective function, as shown in formula (22):

[0063] C SES =min(C INV +C OM -C SES,DN ) (twenty two);

[0064] a) Investment cost of shared energy storage C INV , as shown in formula (23):

[0065]

[0066] b) The operation and maintenance cost of shared energy storage, as shown in formula (24):

[0067]

[0068] c) the cost of electricity interaction with the distribution network, as shown in the above formula (14);

[0069] (b) Constraints

[0070] a) Shared energy storage constraints, as shown in formula (25):

[0071]

[0072] b) Shared energy storage power balance constraint, as shown in formula (26):

[0073]

[0074] Where: U buy,SES (t) is 1 when it means buying electricity, and 0 when it means selling electricity; U sell,SES (t) 0 indicates electricity purchase, and 1 indicates electricity sale;

[0075] 2) Establish a multi-microgrid dispatch model

[0076] (a) Objective function, as shown in formula (27):

[0077] C MMG =min(C TU -C MMG,DN ) (27);

[0078] a) Start-up and shutdown and coal consumption costs of small coal-fired units, as shown in formula (28):

[0079]

[0080] Where: P TU,i (t) is the dispatch output of small and medium-sized coal-fired units in microgrid i during period t; a1, b1, c1 are the coal consumption coefficients of small coal-fired units respectively; U TU,i (t) = 1 means start, U TU,i (t) = 0 means shutdown; δ TU is the start-up and shutdown cost coefficient of the unit;

[0081] b) the cost of electricity interaction with the distribution network, as shown in the above formula (13);

[0082] (b) Constraints

[0083] a) Operation constraints of small coal-fired units, as shown in formula (29):

[0084]

[0085] b) Constraints on inter-microgrid electricity trading, as shown in formula (30):

[0086]

[0087] Where: The upper limit of the interactive power between microgrid i and other microgrids;

[0088] c) Microgrid power balance constraint, as shown in formula (31):

[0089]

[0090] Where: P ij (t) is the interaction power between microgrids i and j in time period t, P ij (t)>0 means that microgrid i sells electricity to microgrid j, U buy,i (t) is 1 when it means buying electricity, and 0 when it means selling electricity; U sell,i (t) 0 indicates electricity purchase, and 1 indicates electricity sale;

[0091] d) Tie line power constraint, as shown in formula (32):

[0092]

[0093] The beneficial effects of the present invention are:

[0094] (1) By constructing a multi-objective optimization configuration model and optimization algorithm, a balanced consideration of all objectives is achieved, effectively smoothing the fluctuations of each microgrid's access to renewable energy and reducing the shared energy storage costs that each microgrid needs to bear;

[0095] (2) By constructing a multi-agent game dispatching framework with one master and two slaves, it is conducive to promoting cooperation and coordination among the distribution network, shared energy storage, and multiple microgrids, realizing the coordination of the three-party dispatching, and then achieving a balanced distribution of the interests of the distribution network, shared energy storage, and multiple microgrids, thereby improving the overall benefits. BRIEF DESCRIPTION OF THE DRAWINGS

[0096] Figure 1 This is the process of the distribution network-shared energy storage-multi-microgrid optimization scheduling method based on multi-agent game of the present invention;

[0097] Figure 2 It is a flowchart for solving the game optimization model in the distribution network-shared energy storage-multi-microgrid optimization scheduling method based on multi-agent game of the present invention;

[0098] Figure 3 It is a net load fluctuation curve diagram of the microgrid 1 of the present invention before and after the shared energy storage is charged and discharged;

[0099] Figure 4 It is a net load fluctuation curve diagram of the microgrid 2 of the present invention before and after the shared energy storage is charged and discharged;

[0100] Figure 5 It is a net load fluctuation curve diagram of the microgrid 3 of the present invention before and after the shared energy storage is charged and discharged. DETAILED DESCRIPTION

[0101] The present invention is described in detail below with reference to the accompanying drawings and specific embodiments.

[0102] The present invention provides a distribution network-shared energy storage-multi-microgrid optimization scheduling method based on multi-agent game, such as Figure 1 As shown, the specific implementation steps are as follows:

[0103] Step 1: Construct a multi-objective optimization configuration model for multi-microgrid shared energy storage that takes into account energy storage investment costs and load fluctuations. Specifically:

[0104] Step 1.1, establish two optimization objectives: minimum energy storage investment cost and minimum net load standard deviation;

[0105] (1) The energy storage cost of microgrid i is the lowest, as shown in formula (1):

[0106] C SES,i =min(C INV,i +C OM,i ) (1);

[0107] Where: C SES.i is the energy storage investment cost of microgrid i, C INV..i is the energy storage investment cost borne by microgrid i, C OM.i Energy storage operation and maintenance costs borne by microgrid i;

[0108] The cost calculation formulas are as follows:

[0109] 1) Without considering the impact of economic development and inflation, the energy storage investment cost borne by microgrid i is shown in formula (2):

[0110]

[0111] Where: P , E They represent the unit power investment cost and unit capacity investment cost of shared energy storage respectively; They represent the maximum values ​​of energy storage power and capacity required by microgrid i; T E Represents the service life of the shared energy storage.

[0112] 2) The energy storage operation and maintenance cost borne by microgrid i is shown in formula (3):

[0113]

[0114] Where: P SES,C,i (t), P SES,D,i (t) respectively represent the charging and discharging power of energy storage to microgrid i in time period t; δ OM The unit power operation and maintenance cost of energy storage for microgrid i; t = 1, 2, 3…T, T is the scheduling period; Δt is the time interval;

[0115] (2) The net load standard deviation of microgrid i is the smallest, as shown in equations (4)-(5):

[0116]

[0117] Where: S i is the net load standard deviation of microgrid i; P load,i (t) is the load value of microgrid i in period t; P WT,i (t), P PV,i (t) are the wind power and photovoltaic power absorbed by microgrid i in period t; is the average equivalent load of microgrid i;

[0118] Step 1.2, determine the constraints of the two optimization objectives, and optimize the shared energy storage for the charging and discharging power of each microgrid; the constraints include the coupling constraints and the charging and discharging power constraints shared with microgrid i;

[0119] (1) The coupling constraint shared with microgrid i is shown in formula (6):

[0120]

[0121] is the upper limit of the energy storage power demand of microgrid i, as shown in formula (7):

[0122]

[0123] Where: P SES,C,i (t), P SES,D,i (t) represents the charging and discharging power of energy storage to microgrid i in time period t; They represent the upper limits of charging and discharging power of energy storage to microgrid i; U SES,C,i (t), U SES,D,i (t) Ensure that the energy storage can only charge or discharge microgrid i at the same time during the dispatch cycle, U SES,C,i (t) 1 indicates charging, and 0 indicates discharging;

[0124] (2) Charge and discharge power constraints, as shown in equations (8)-(10):

[0125] E i (t) = E i (t-1)+(η SES,C P SES,C,i (t)-P SES,D,i (t) / η SES,D )Δt (8);

[0126]

[0127] Where: η SES,C , η SES,D Respectively represent the charging and discharging efficiency of energy storage; E i (t) represents the storage capacity required for energy storage to charge and discharge microgrid i during time period t; is the upper limit of the capacity demand of microgrid i using energy storage, μ E It is the capacity margin of microgrid shared energy storage.

[0128] Step 2: Use the multi-objective particle swarm algorithm to solve the multi-objective optimization configuration model of multi-microgrid shared energy storage in step 1, as shown in Figure 2 As shown, the charging and discharging power of energy storage to each microgrid is optimized; specifically:

[0129] Step 2.1, initialize N particle positions, velocities, pbest and gbest, and the number of iterations inter = 0;

[0130] Step 2.2, calculate the fitness values ​​of the initialized N populations, i.e., the energy storage investment cost and the mean square error of the net load of the multi-microgrid;

[0131] Step 2.3, update the particle's velocity and position, set the dynamic inertia weight, and update the population individual extreme value pbest;

[0132] Step 2.4, find the Pareto optimal solution of the new particle, merge it into the external archive set, find the Pareto optimal solution in the external archive set, and remove the non-Pareto optimal solution;

[0133] Step 2.5: Determine whether the number of optimal solutions in the external set exceeds the specified capacity m. If so, select m particles based on the crowding distance. Otherwise, use the roulette method to select gbest.

[0134] Step 2.6, the particle mutates with a small probability of 30% within the original position;

[0135] Step 2.7, determine whether inter=inter+1 and inter=intermax are satisfied. If so, output the Pareto optimal solution set and record the optimal charging and discharging strategy of each microgrid. Otherwise, return to step 2.3;

[0136] Step 3: Establish a one-master-two-slave game dispatching model with the distribution network as the leader and multiple microgrids and shared energy storage as followers, so as to achieve a balanced distribution of the interests of the distribution network, shared energy storage and multiple microgrids, thereby improving the overall benefits; specifically:

[0137] (1) Establishing the main distribution network dispatching model

[0138] 1) Objective function

[0139] The decision variable for optimal dispatching of distribution network is the purchase and sale price of electricity, and the goal of optimal operation is the total cost C DN Minimum, as shown in formula (11):

[0140] C DN =min(C grid +C MMG,DN +C SES,DN ) (11);

[0141] Where: C DN is the total cost of the distribution network; C grid The cost of purchasing electricity from the main grid for the distribution network; C MMG,DN is the energy interaction cost between the distribution network and the multi-microgrid; C SES,DN is the energy interaction cost between the distribution network and shared energy storage. The cost calculation formula is as follows:

[0142] (a) The cost of purchasing electricity from the main grid is shown in formula (12):

[0143]

[0144] Where: sell,grid (t) are the electricity prices of the main grid in time period t; P buy,DN (t) is the power purchased by the distribution network from the main grid during time period t;

[0145] (b) The cost of electricity interaction with multiple microgrids, as shown in formula (13):

[0146]

[0147] Where: bu、y,DN (t) is the electricity purchase price of the optimized distribution network from the microgrid and shared energy storage in time period t; sell,DN (t) is the electricity price sold by the optimized distribution network to the microgrid and shared energy storage; P buy,i (t), P sell,i (t) is the power purchased and sold by microgrid i from the distribution network during time period t.

[0148] (c) The cost of electricity interaction with shared energy storage, as shown in formula (14):

[0149]

[0150] Where: P buy,SES (t), P sell,SES (t) is the power purchased and sold by the shared energy storage from the distribution network during period t.

[0151] 2) Constraints

[0152] The constraints include power flow balance constraints, state variable constraints, distributed energy access constraints, and time-of-use electricity price constraints;

[0153] (a) Power flow balance constraint, as shown in formula (15):

[0154]

[0155] Where: n is the number of distribution network nodes; P p , Q p , U p are the active power, reactive power and voltage amplitude injected into node p respectively; G pq , B pq , δ pq are the conductance, susceptance and voltage phase difference between nodes p and q respectively. q is the voltage amplitude at node q;

[0156] (b) State variable constraints, as shown in formula (16):

[0157]

[0158] Where: are the upper and lower limits of the voltage amplitude injected into node p respectively; G pq , B pq , δ pq are the conductance, susceptance and voltage phase difference between nodes p and q respectively; Sl , are the power and power upper limit of distribution network branch l respectively.

[0159] (c) Distributed energy access constraints, as shown in formula (17):

[0160]

[0161] Where: P DG (t), Q DG (t) are respectively the active and reactive outputs of distributed energy in time period t; They are the upper and lower limits of the active output of distributed energy in time period t; They are the upper and lower limits of the reactive power output of distributed energy in time period t respectively.

[0162] (d) Time-of-use electricity price constraints;

[0163] The electricity purchase and sale prices set by the distribution network for microgrids and shared energy storage are time-of-use electricity prices. The electricity sales prices and electricity purchase prices during off-peak hours, normal hours, and peak hours satisfy the following relationship, as shown in equations (18)-(19):

[0164]

[0165] Where: T v , T f , T p are valley period, normal period and peak period respectively; δ buyv,DN (t),δ buyf,DN (t) and δ buyp,DN (t) are the power purchase prices of the distribution network during off-peak hours, normal hours and peak hours respectively; δ sellv,DN (t),δ sellf,DN (t) and δ sellp,DN (t) are the electricity prices of the distribution network during off-peak, normal and peak periods respectively;

[0166] In order to ensure the profitability of the distribution network, the electricity sales price of the distribution network in each period must be higher than the electricity purchase price, as shown in formula (20):

[0167]

[0168] In order to prevent the main distribution network from raising the electricity price for its own interests during the game process, the average electricity price within the dispatch period T should not be greater than the upper limit of the distribution network electricity price, as shown in formula (21):

[0169]

[0170] Where: It is the upper limit of the electricity price of the distribution network.

[0171] (2) Establish a multi-microgrid dispatching model with shared energy storage;

[0172] 1) Establish a shared energy storage dispatch model

[0173] (a) Objective function

[0174] The charging and discharging power P of the shared energy storage to the microgrid i has been optimized by step 2. SES,C,i (t), P SES,D,i (t), so the decision variable of the shared energy storage scheduling model is the interaction power with the distribution network, and the goal of the optimized operation is the total cost C SES Minimum, as shown in formula (22):

[0175] C SES =min(C INV +C OM -C SES,DN ) (twenty two);

[0176] Where: C SES The total cost of shared energy storage; C INV is the investment cost of shared energy storage; C OM is the operation and maintenance cost of shared energy storage; C SES,DN is the electricity interaction cost between shared energy storage and distribution network. The calculation formulas for each cost are as follows:

[0177] a) Investment cost of shared energy storage C INV , as shown in formula (23):

[0178]

[0179] Where: P and λ E They are the unit power investment cost and unit capacity investment cost of shared energy storage respectively; and are the upper power limit and capacity limit of shared energy storage respectively; T E The useful life of shared energy storage;

[0180] b) The operation and maintenance cost of shared energy storage, as shown in formula (24):

[0181]

[0182] Where: P SES,C (t), P SES,D (t) represents the charging and discharging power of the energy storage in time period t; δ OM is the unit power operation and maintenance cost of energy storage;

[0183] c) the cost of electricity interaction with the distribution network, as shown in the above formula (14);

[0184] (b) Constraints

[0185] a) Shared energy storage constraints include the inherent constraints of shared energy storage charging and discharging, upper and lower limit constraints of power consumption, and the power relationship constraints between two adjacent time periods, as shown in formula (25):

[0186]

[0187] Where: E SES (t) represents the capacity of shared energy storage in time period t; η SES,C , η dSES,D Respectively represent the charging and discharging efficiency of energy storage; P SES,C (t), P SES,D (t) represents the charging and discharging power of the energy storage within the time period t; E SES (0) and E SES (24) represent the energy stored at the start and end of the dispatch period respectively; Indicates the upper limit value of the energy storage capacity of the energy storage; Indicates the upper limit of the charging and discharging power of energy storage; U SES,C (t), U SES,D (t) Ensure that the energy storage can only be charged or discharged at the same time during the dispatch cycle, U SES,C When (t) is 1, it indicates charging, and when it is 0, it indicates discharging.

[0188] b) Shared energy storage power balance constraint, as shown in formula (26):

[0189] Here, the charging and discharging power P of the shared energy storage optimized in step 2 to the microgrid i is used. SES,C,i (t), P SES,D,i (t);

[0190]

[0191] Where: U buy,SES (t), U sell,SES (t) Ensure that the energy storage can only purchase or sell electricity with the distribution network at the same time during the dispatch cycle, U buy,SES When (t) is 1, it indicates purchasing electricity, and when it is 0, it indicates selling electricity. buy,SES (t), P sell,SES (t) is the power purchased and sold by the shared energy storage from the distribution network during time period t.

[0192] 2) Establish a multi-microgrid dispatch model

[0193] (a) Objective function, as shown in formula (27):

[0194] C MMG =min(CTU -C MMG,DN ) (27);

[0195] Where: C MMG is the total cost of the multi-microgrid; C TU is the start-up and shutdown and coal consumption cost of small coal-fired units; C MMG,DN is the energy interaction cost between multi-microgrids and distribution network. The cost calculation formula is as follows:

[0196] a) Start-up and shutdown and coal consumption costs of small coal-fired units, as shown in formula (28):

[0197]

[0198] Where: P TU,i (t) is the dispatch output of small and medium-sized coal-fired units in microgrid i during period t; a1, b1, c1 are the coal consumption coefficients of small coal-fired units respectively; U TU,i (t) is a binary variable, U TU,i (t) = 1 means start, U TU,i (t) = 0 means shutdown; δ TU is the start-up and shutdown cost coefficient of the unit; i=1,2,3…N, N is a positive integer;

[0199] b) the cost of electricity interaction with the distribution network, as shown in the above formula (13);

[0200] (b) Constraints

[0201] a) Operation constraints of small coal-fired units, as shown in formula (29):

[0202]

[0203] Where: They are the upper and lower limits of the output power of small coal-fired units respectively; They are respectively the upper and lower limits of the output power ramp of small coal-fired units.

[0204] b) Constraints on inter-microgrid electricity trading, as shown in formula (30):

[0205]

[0206] Where: is the upper limit of the interactive power between microgrid i and other microgrids.

[0207] c) Microgrid power balance constraint, as shown in formula (31):

[0208]

[0209] Where: P ij(t) is the interaction power between microgrids i and j in time period t, P ij (t)>0 means that microgrid i sells electricity to microgrid j, U buy,i (t), U sell,i (t) Ensure that the microgrid can only interact with the distribution network at the same time during the dispatch cycle, U buy,i When (t) is 1, it indicates purchasing electricity, and when it is 0, it indicates selling electricity.

[0210] d) Tie line power constraint, as shown in formula (32):

[0211]

[0212] Where: is the maximum transmission power of the tie line;

[0213] Step 4: On the basis of optimizing the charging and discharging power of energy storage to each microgrid, a genetic algorithm nested solver is used to solve the one-master-two-slave game scheduling model to obtain the optimal scheduling solution; specifically:

[0214] Step 4.1, initialize (update) genetic population parameters, generate (update) distribution network time-of-use electricity price;

[0215] Step 4.2, use Gurobi to solve the minimum cost of shared energy storage and multi-microgrids respectively, and record the optimal power purchase and sales strategy between the distribution network and shared energy storage and the optimal power purchase and sales strategy between the distribution network and multi-microgrids;

[0216] Step 4.3, calculate the lowest cost of the distribution network of the global optimal genetic population and record the electricity price;

[0217] Step 4.4: Determine whether it is lower than the previously recorded cost. If so, update the optimal cost of the distribution network. Otherwise, return to step 4.1.

[0218] Step 4.5: Determine whether the iteration limit is reached. If so, output the optimal scheduling strategy for each subject. Otherwise, return to step 4.1.

[0219] Step 4.6: Allocate the cost among multiple microgrids based on the Shapley value. The specific calculation process is as follows:

[0220] The Shapley value method is a commonly used cost allocation method. If the total cost C of the multi-microgrid is calculated based on the conventional Shapley value method, MMG The cost shared by microgrid i can be expressed as formula (33):

[0221]

[0222] Where: N is the number of microgrids; S i represents the set consisting of all subsets of microgrid i; Si ={1,2,…,N}; |s| represents the number of microgrids contained in alliance s; v(s) represents the cost generated by alliance s; v(s\{i}) represents the cost generated by alliance s after removing microgrid i.

[0223] The present invention adopts an improved Shapley value method, that is, using the interaction power P between microgrid i and microgrid j ij (t) Measure the contribution to the alliance and redefine the share of microgrid i i , as shown in formula (34):

[0224]

[0225] Among them, the sum of the proportions of N microgrids satisfies The difference between the conventional specific gravity and the improved specific gravity of microgrid i, ΔB i =B i -1 / N, using ΔB i The allocation result of the conventional Shapley value method can be corrected by the correction amount ΔB i , then the modified shared cost of microgrid i is as shown in formula (35):

[0226]

[0227] Example 1

[0228] In order to verify the effectiveness of the model and method, the present invention uses the improved IEEE33 node system for simulation testing. Node 1 of the distribution network is connected to the upper power grid and is equipped with a wind farm and a photovoltaic power station, which are connected to nodes 19 and 9 respectively. The three microgrids MG1, MG2 and MG3 are connected to nodes 29, 32 and 15 respectively, and the shared energy storage is connected to node 13. Table 1 shows the parameters of the distribution network and shared energy storage; in the simulation process, the population number of the genetic algorithm is set to 20 and the number of iterations is set to 100.

[0229] Table 1 Distribution network and shared energy storage parameters

[0230]

[0231]

[0232] Table 2 shows the time-of-use electricity prices set by the distribution network. The prices in the valley, flat and peak periods show a gradual upward trend. In each specific time period, the electricity price sold by the distribution network to users is always higher than the price they purchase electricity. This pricing mechanism effectively ensures the economically stable operation of the distribution network.

[0233] Table 2 Time-of-use electricity prices set by distribution networks

[0234]

[0235] Example 2

[0236] Figure 3-Figure 5 The net load fluctuation diagram of each microgrid before and after the shared energy storage is charged and discharged. The black column represents the charging and discharging power of the energy storage in microgrid i. A value greater than 0 represents energy storage charging and an increase in the net load of the microgrid. A value less than 0 represents energy storage discharging and a decrease in the net load of the microgrid. It can be seen that the established model fully stimulates the role of energy storage and effectively smoothes the net load fluctuation caused by the anti-peaking characteristics of new energy. This mechanism helps to balance the supply and demand differences of the power grid and significantly improve the stability and reliability of the power grid.

[0237] Example 3

[0238] In order to verify the effectiveness of the proposed one-master-two-slave game model, the distribution network is taken as the leader, and whether shared energy storage and multi-microgrids respond to the time-of-use electricity price set by the distribution network as followers is considered. Three operation schemes are set for comparative analysis, as shown in Table 3; based on the solution results, the three-party costs of the three operation schemes are compared, as shown in Table 4.

[0239] Table 3 Comparison of three operation schemes

[0240] Operation plan Is shared energy storage a follower? Whether Multi-Micronet acts as a follower 1 yes yes 2 no yes 3 no no

[0241] Table 4 Comparison of three-party costs of three operation schemes

[0242] Operation plan Distribution network cost / yuan Shared energy storage cost / yuan Multi-microgrid cost / yuan 1 59624 2173 8664 2 60213 2456 8672 3 63452 3863 10457

[0243] In Scheme 2, the distribution network and multi-microgrids form a one-master-one-slave game. Compared with Scheme 1, the shared storage energy cannot respond to the time-of-use electricity price set by the distribution network, and cannot use its remaining capacity after charging and discharging to the multi-microgrid. Therefore, the cost of the distribution network and shared energy storage has increased, and the cost of the multi-microgrid remains almost unchanged; in Scheme 3, the shared energy storage and multi-microgrids are dispatched separately according to the fixed electricity price. Compared with Scheme 1, the shared energy storage and multi-microgrids cannot respond to the time-of-use electricity price set by the distribution network, and can only adjust their own power purchase and sales strategies to the distribution network, that is, purchase electricity when the electricity price is low and sell electricity when the electricity price is high. Therefore, the cost of all three parties increases. Therefore, it is proved that the distribution network in Scheme 1 is the leader, and the shared energy storage and multi-microgrids are followers, forming a one-master-two-slave game strategy is the most reasonable.

[0244] Example 4

[0245] The effectiveness analysis of the improved Shapley value method is as follows:

[0246] The cost comparison of the microgrid before and after joining the alliance is shown in Table 5. The total operating cost after the establishment of the microgrid alliance is 8664.06 yuan, which is a 3.6% decrease compared with the total cost of independent operation of each microgrid before the alliance. At the same time, the cost shared by each microgrid after the alliance is also reduced compared with independent operation, which fully proves that the alliance has reached a game equilibrium and achieved the optimization of economic benefits.

[0247] Table 5 Comparison of costs before and after microgrids join the alliance

[0248] Before joining the alliance After joining the alliance Microgrid 1 2318.09 2149.63 Microgrid 2 3052.88 3021.97 Microgrid 3 3625.92 3492.40 total 8996.89 8664.06

[0249] The comparison between the traditional Shapley algorithm and the improved Shapley algorithm is shown in Table 6. It can be seen that compared with the traditional Shapley algorithm, the improved Shapley algorithm measures the contribution to the alliance by the interaction power between microgrids. Since the interaction power between microgrid 2 and other microgrids is low, the cost of microgrid 2 increases by 14.7%. The interaction power between microgrids 1 and 3 and other microgrids is high, so the costs of microgrid 1 and microgrid 3 are reduced by 12.0% and 5.06% respectively, and the overall cost of multi-microgrids is reduced by 1.06%. This not only reflects the advantages of individual microgrids in the alliance, but also provides an effective idea for the problem of multi-microgrid benefit distribution.

[0250] Table 6 Comparison between traditional Shapley algorithm and improved Shapley algorithm

[0251]

Claims

1. A distribution network-shared energy storage-multi-microgrid optimization scheduling method based on multi-agent game, characterized in that: Follow the steps below to implement it: Step 1: Construct a multi-objective optimization configuration model for multi-microgrid shared energy storage that takes into account energy storage investment costs and load fluctuations; Step 2, using a multi-objective particle swarm algorithm to solve the multi-objective optimization configuration model of multi-microgrid shared energy storage in step 1, and optimize the charging and discharging power of energy storage to each microgrid; Step 3: Establish a one-master-two-slave game dispatching model with the distribution network as the leader and multiple microgrids and shared energy storage as followers; Step 4: Based on the optimization of the charging and discharging power of the energy storage to each microgrid, the genetic algorithm nested solver is used to solve the one-master-two-slave game scheduling model to obtain the optimal scheduling solution.

2. The distribution network-shared energy storage-multi-microgrid optimization scheduling method based on multi-agent game as claimed in claim 1 is characterized in that: In the step 1, specifically: Step 1.1, establish two optimization objectives: minimum energy storage investment cost and minimum net load standard deviation; (1) The energy storage cost of microgrid i is the lowest, as shown in formula (1): C SES,i =min(C INV,i +C OM,i ) (1); Where: C SES.i is the energy storage investment cost of microgrid i, C INV.i is the energy storage investment cost borne by microgrid i, C OM.i Energy storage operation and maintenance costs borne by microgrid i; The cost calculation formulas are as follows: 1) The energy storage investment cost borne by microgrid i is shown in formula (2): Where: P , E They represent the unit power investment cost and unit capacity investment cost of shared energy storage respectively; They represent the maximum values ​​of energy storage power and capacity required by microgrid i; T E represents the service life of the shared energy storage; 2) The energy storage operation and maintenance cost borne by microgrid i is shown in formula (3): Where: P SES,C,i (t), P SES,D,i (t) respectively represent the charging and discharging power of energy storage to microgrid i in time period t; δ OM The unit power operation and maintenance cost of energy storage for microgrid i; T is the dispatch period; Δt is the time interval; (2) The net load standard deviation of microgrid i is the smallest, as shown in equations (4)-(5): Where: S i is the net load standard deviation of microgrid i; P load,i (t) is the load value of microgrid i in period t; P WT,i (t), P PV,i (t) are the wind power and photovoltaic power absorbed by microgrid i in period t; is the average equivalent load of microgrid i; Step 1.2, determine the constraints of the two optimization objectives, and optimize the shared energy storage for the charging and discharging power of each microgrid; the constraints include the coupling constraints and the charging and discharging power constraints shared with microgrid i; (1) The coupling constraint shared with microgrid i is shown in formula (6): is the upper limit of the energy storage power demand of microgrid i, as shown in formula (7): Where: They represent the upper limits of the charging and discharging power of energy storage to microgrid i; U SES,C,i (t) is 1 for charging and 0 for discharging; U SES,D,i (t) 1 indicates discharging, and 0 indicates charging; (2) Charge and discharge power constraints, as shown in equations (8)-(10): From i (t)=E i (t-1)+(η SES,C P SES,C,i (t)-P SES,D,i (t) / η SES,D )Δt (8); Where: η SES,C , η SES,D Respectively represent the charging and discharging efficiency of energy storage; E i (t) represents the storage capacity required for energy storage to charge and discharge microgrid i during time period t; is the upper limit of the capacity demand of microgrid i using energy storage, μ E It is the capacity margin of microgrid shared energy storage.

3. The distribution network-shared energy storage-multi-microgrid optimization scheduling method based on multi-agent game as claimed in claim 2 is characterized in that: In the step 2, specifically: Step 2.1, initialize N particle positions, velocities, pbest and gbest, and the number of iterations inter = 0; Step 2.2, calculate the fitness values ​​of the initialized N populations, i.e., the energy storage investment cost and the mean square error of the net load of the multi-microgrid; Step 2.3, update the particle's velocity and position, set the dynamic inertia weight, and update the population individual extreme value pbest; Step 2.4, find the Pareto optimal solution of the new particle, merge it into the external archive set, find the Pareto optimal solution in the external archive set, and remove the non-Pareto optimal solution; Step 2.5: Determine whether the number of optimal solutions in the external set exceeds the specified capacity m. If so, select m particles based on the crowding distance. Otherwise, use the roulette method to select gbest. Step 2.6, the particle mutates with a small probability of 30% within the original position; Step 2.7: Determine whether inter=inter+1 and inter=intermax are satisfied. If so, output the Pareto optimal solution set and record the optimal charging and discharging strategy of each microgrid. Otherwise, return to step 2.

3.

4. The distribution network-shared energy storage-multi-microgrid optimization scheduling method based on multi-agent game as claimed in claim 3 is characterized in that: In the step 3, specifically: (1) Establishing the main distribution network dispatching model 1) Objective function The decision variable for optimal dispatching of distribution network is the purchase and sale price of electricity, and the goal of optimal operation is the total cost C DN Minimum, as shown in formula (11): C DN =min(C grid +C MMG,DN +C SES,DN ) (11); Where: C DN is the total cost of the distribution network; C grid The cost of purchasing electricity from the main grid for the distribution network; C MMG,DN is the energy interaction cost between the distribution network and the multi-microgrid; C SES,DN is the energy interaction cost between the distribution network and shared energy storage; the cost calculation formula is as follows: (a) The cost of purchasing electricity from the main grid is shown in formula (12): Where: sell,grid (t) are the electricity prices of the main grid in period t; P buy,DN (t) is the power purchased by the distribution network from the main grid during time period t; (b) The cost of electricity interaction with multiple microgrids, as shown in formula (13): Where: bu、y,DN (t) is the electricity purchase price of the optimized distribution network from the microgrid and shared energy storage in time period t; sell,DN (t) is the electricity price sold by the optimized distribution network to the microgrid and shared energy storage; P buy,i (t), P sell,i (t) is the power purchased and sold by microgrid i from the distribution network in time period t; (c) The cost of electricity interaction with shared energy storage, as shown in formula (14): Where: P buy,SES (t), P sell,SES (t) is the power purchased and sold by the shared energy storage from the distribution network during period t; 2) Constraints Constraints include power flow balance constraints, state variable constraints, distributed energy access constraints, and time-of-use electricity price constraints; (a) Power flow balance constraint, as shown in formula (15): Where: n is the number of distribution network nodes; P p , Q p , U p are the active power, reactive power and voltage amplitude injected into node p respectively; G pq , B pq ,δ pq are the conductance, susceptance and voltage phase difference between nodes p and q respectively; U q is the voltage amplitude at node q; (b) State variable constraints, as shown in formula (16): Where: are the upper and lower limits of the voltage amplitude injected into node p respectively; G pq , B pq ,δ pq are the conductance, susceptance and voltage phase difference between nodes p and q respectively; S l , are the power and power upper limit of distribution network branch l respectively; (c) Distributed energy access constraints, as shown in formula (17): Where: P DG (t), Q DG (t) are respectively the active and reactive outputs of distributed energy in time period t; They are the upper and lower limits of the active output of distributed energy in time period t; They are the upper and lower limits of the reactive power output of distributed energy in time period t respectively; (d) Time-of-use electricity price constraints; The electricity purchase and sale prices set by the distribution network for microgrids and shared energy storage are time-of-use electricity prices. The electricity sales prices and electricity purchase prices during off-peak hours, normal hours, and peak hours satisfy the following relationship, as shown in equations (18)-(19): Where: T v 、T f 、T p are valley period, normal period and peak period respectively; δ buyv,DN (t),δ buyf,DN (t) and δ buyp,DN (t) are the power purchase prices of the distribution network during off-peak hours, normal hours and peak hours respectively; δ sellv,DN (t),δ sellf,DN (t) and δ sellp,DN (t) are the electricity prices of the distribution network during off-peak, normal and peak periods respectively; The electricity selling price of the distribution network in each period must be higher than the electricity purchasing price, as shown in formula (20): The average electricity price within the dispatch period T should not be greater than the upper limit of the distribution network electricity price, as shown in formula (21): Where: The upper limit of the electricity price sold by the distribution network; (2) Establish a multi-microgrid dispatching model with shared energy storage; 1) Establish a shared energy storage dispatch model (a) Objective function The charging and discharging power P of the shared energy storage to the microgrid i has been optimized by step 2. SES,C,i (t), P SES,D,i (t), so the decision variable of the shared energy storage scheduling model is the interaction power with the distribution network, and the goal of the optimized operation is the total cost C SES Minimum, as shown in formula (22): C SES =min(C INV +C OM -C SES,DN ) (22); Where: C SES The total cost of shared energy storage; C INV is the investment cost of shared energy storage; C OM is the operation and maintenance cost of shared energy storage; the cost calculation formula is as follows: a) Investment cost of shared energy storage C INV , as shown in formula (23): Where: P and λ E They are the unit power investment cost and unit capacity investment cost of shared energy storage respectively; and are the upper power limit and capacity limit of shared energy storage respectively; T E The useful life of shared energy storage; b) The operation and maintenance cost of shared energy storage, as shown in formula (24): In the formula, δ OM is the unit power operation and maintenance cost of energy storage; c) the cost of electricity interaction with the distribution network, as shown in the above formula (14); (b) Constraints a) Shared energy storage constraints, as shown in formula (25): Where: E SES (t) represents the capacity of shared energy storage in time period t; η SES,C , η dSES,D Respectively represent the charging and discharging efficiency of energy storage E SES (0) and E SES (24) represent the energy stored at the start and end of the dispatch period respectively; b) Shared energy storage power balance constraint, as shown in formula (26): Where: U buy,SES (t) is 1 when it means buying electricity, and 0 when it means selling electricity; U sell,SES (t) 0 indicates electricity purchase, and 1 indicates electricity sale; 2) Establish a multi-microgrid dispatch model (a) Objective function, as shown in formula (27): C MMG =min(C TU -C MMG,DN ) (27); Where: C MMG is the total cost of multi-microgrid; C TU is the start-up and shutdown and coal consumption cost of small coal-fired units; C MMG,DN is the electric energy interaction cost between multi-microgrids and distribution network; the calculation formulas of various costs are as follows: a) Start-up and shutdown and coal consumption costs of small coal-fired units, as shown in formula (28): Where: P TU,i (t) is the dispatch output of small and medium-sized coal-fired units in microgrid i during period t; a1, b1, c1 are the coal consumption coefficients of small coal-fired units respectively; U TU,i (t) = 1 means start, U TU,i (t) = 0 means shutdown; δ TU is the start-up and shutdown cost coefficient of the unit; b) the cost of electricity interaction with the distribution network, as shown in the above formula (13); (b) Constraints a) Operation constraints of small coal-fired units, as shown in formula (29): Where: They are the upper and lower limits of the output power of small coal-fired units respectively; They are the upper and lower limits of the output power ramp of small coal-fired units respectively; b) Constraints on inter-microgrid electricity trading, as shown in formula (30): Where: The upper limit of the interactive power between microgrid i and other microgrids; c) Microgrid power balance constraint, as shown in formula (31): Where: P ij (t) is the interaction power between microgrids i and j in time period t, P ij (t)>0 means that microgrid i sells electricity to microgrid j, U buy,i (t) is 1 when it means buying electricity, and 0 when it means selling electricity; U sell,i (t) 0 indicates electricity purchase, and 1 indicates electricity sale; d) Tie line power constraint, as shown in formula (32): Where: is the maximum transmission power of the interconnection line.

5. The distribution network-shared energy storage-multi-microgrid optimization scheduling method based on multi-agent game as claimed in claim 4 is characterized in that: In the step 4, specifically: Step 4.1, initialize genetic population parameters and generate time-of-use electricity prices for distribution networks; Step 4.2, use Gurobi to solve the minimum cost of shared energy storage and multi-microgrids respectively, and record the optimal power purchase and sales strategy between the distribution network and shared energy storage and the optimal power purchase and sales strategy between the distribution network and multi-microgrids; Step 4.3, calculate the lowest cost of the distribution network of the global optimal genetic population and record the electricity price; Step 4.4: Determine whether it is lower than the previously recorded cost. If so, update the optimal cost of the distribution network. Otherwise, return to step 4.

1. Step 4.5: Determine whether the iteration limit is reached. If so, output the optimal scheduling strategy for each subject. Otherwise, return to step 4.

1. Step 4.6: Allocate the cost among multiple microgrids according to the Shapley value, i.e., use the interaction power P between microgrid i and microgrid j ij (t) Measure the contribution to the alliance and redefine the share of microgrid i i , as shown in formula (34): Among them, the sum of the proportions of N microgrids satisfies The difference between the conventional specific gravity and the improved specific gravity of microgrid i, ΔB i =B i -1 / N, using ΔB i Correct the allocation result by ΔB i , then the modified shared cost of microgrid i is as shown in formula (35):

Images