Source network load storage collaborative planning method and system considering multi-agent mixed game

Abstract

The invention relates to a source network load storage collaborative planning method and system considering a multi-agent mixed game, and the method comprises the following steps: building a power distribution network collaborative planning model considering a source network load storage agent master-slave-cooperation mixed game, and enabling the power distribution network collaborative planning model to be a double-layer model, comprising an upper-layer master-slave game sub-model of a power distribution network operator and other subjects and a lower-layer cooperative game sub-model between a source load storage side subject; and solving the power distribution network collaborative planning model by adopting a solving strategy combining a target cascade analysis method and an adaptive coefficient alternating direction multiplier method to obtain an optimal planning scheme of a new line, a distributed power supply and an energy storage device. Compared with the prior art, the method has the advantages that efficient configuration and coordinated utilization of source network load storage resources can be realized, and the planning reliability is improved.

Description

Technical Field

[0001] The present invention belongs to the field of distribution networks, and relates to a distribution network planning method, and in particular to a source-grid-load-storage collaborative planning method and system considering multi-agent hybrid game. Background Art

[0002] With the deepening of the market-oriented reform of electricity, various types of social capital have joined the construction of distribution networks, presenting a situation of competition among stakeholders. The coordinated and rapid development of the four aspects of the new distribution system, "source, grid, load, and storage", has increased the number of various types of elements in the distribution system, increased the number of operation and maintenance entities, deepened the interaction between sources and loads, and made the operation characteristics more complex. Each entity has its own interests and demands, and also needs to undertake the important task of maintaining the efficient operation of the system. Diversified entities and competitive environment have put forward higher requirements for distribution network planning. The actual need to take into account the specific interests of different entities has brought new challenges to the refined planning of the distribution system. Specifically, the construction of distributed energy stations is likely to lead to an increase in peak loads and line load rates exceeding safety standards. The access of distributed photovoltaics leads to an increase in the peak-to-valley difference rate of line currents, and energy storage devices need to be connected to improve the system operation level. These problems have led to an increase in the investment cost of power grid planning. In this context, how to consider the interests of various entities and reasonably describe the game situation, and then realize the efficient allocation and coordinated utilization of source, grid, load, and storage resources in planning is an urgent problem to be solved.

[0003] Existing research rarely focuses on the need for independent optimization of each subject, or does not reflect the market-based game relationship between multiple subjects on the source, grid, load and storage side of the distribution network planning, making it difficult to promote the consumption of new energy and the improvement of electricity users' electricity consumption behavior. Summary of the invention

[0004] The purpose of the present invention is to overcome the defects of the above-mentioned prior art and to provide a source-grid-load-storage collaborative planning method and system that is suitable for 110kV and below distribution networks and can achieve efficient configuration and coordinated utilization of source-grid-load-storage resources and consider multi-agent hybrid game.

[0005] The purpose of the present invention can be achieved by the following technical solutions:

[0006] A source-grid-load-storage collaborative planning method considering multi-agent hybrid game includes the following steps:

[0007] A distribution network collaborative planning model considering the master-slave-cooperative hybrid game of source-grid-load-storage subjects is established. The source-grid-load-storage subjects include distribution network operators, distributed power generation operators, energy storage operators and power users. The distributed power generation operators, energy storage operators and power users constitute the source-load-storage side subjects. The distribution network collaborative planning model is a two-layer model, including an upper master-slave game sub-model between the distribution network operator and other subjects and a lower cooperative game sub-model between the source-load-storage side subjects.

[0008] The distribution network collaborative planning model is solved by combining the target cascade analysis method with the adaptive coefficient alternating direction multiplier method to obtain the optimal planning scheme for new line construction, distributed power sources and energy storage devices.

[0009] Furthermore, the upper-level master-slave game sub-model is a planning and decision-making model dominated by distribution network operators, with the goal of maximizing the annual comprehensive benefits of distribution network operators. The constraints considered include system flow constraints, system operation safety constraints, new line constraints and transaction power balance constraints.

[0010] Furthermore, the lower-level cooperative game sub-model includes a comprehensive cost minimization sub-problem and a benefit distribution maximization sub-problem, wherein the comprehensive cost minimization sub-problem aims to minimize the sum of the electricity costs of distributed power generation operators, energy storage operators, and power users, and the considered constraints include trading power constraints and electricity trading balance constraints; the benefit distribution maximization sub-problem aims to maximize the incremental benefits of each subject after participating in the cooperation compared to before the cooperation, and the considered constraints include cooperation benefit constraints and trading electricity price constraints.

[0011] Furthermore, the objective function of the benefit distribution maximization sub-problem is expressed as:

[0012]

[0013] In the formula, The benefits after cooperation of the interest subject n; is the benefit of the interest subject n before cooperation; C n is the annual comprehensive cost of entity n; F is the number of entities; α n is the bargaining power of the interest subject n, which is obtained based on the electric energy mapping function.

[0014] Furthermore, when solving the distribution network collaborative planning model, the target cascade analysis method is first used to decouple the two-layer model, and then the adaptive coefficient alternating direction multiplier method is used to distribute the lower-layer cooperative game sub-model.

[0015] Furthermore, in the target cascade analysis method, the end condition of the outer loop iteration is:

[0016]

[0017] In the formula, The power that the stakeholder alliance expects to trade with the distribution network operator; is the power that the distribution network operator expects to trade, with superscripts k and k o is the number of iterations of the internal and external loops of the target cascade analysis method, Ns is the number of scenes, T is the number of time periods, ε o is the convergence threshold.

[0018] Furthermore, in the target cascade analysis method, the penalty factor is updated at the end of the outer loop iteration, and the update formula is:

[0019]

[0020] Where σ is the update constant, The kth o , k o +1 penalty factor for outer loop iteration.

[0021] Furthermore, the distributed solution includes: first solving the comprehensive cost minimization sub-problem to obtain the transaction power between the entities, and then solving the benefit distribution maximization sub-problem to obtain the expected transaction electricity price of the entity.

[0022] Furthermore, the distributed solution specifically includes:

[0023] Establish the augmented Lagrangian function of each sub-problem;

[0024] Based on the expected transaction volume / price information provided by other entities, each entity independently updates its decision and transmits the updated expected transaction volume / price information to other entities;

[0025] Update the Lagrange multiplier and adaptively adjust the penalty factor;

[0026] After the iteration termination condition is met, the final solution is obtained.

[0027] The present invention also provides a source-grid-load-storage collaborative planning system that considers multi-agent hybrid games, comprising one or more processors, a memory, and one or more programs stored in the memory, wherein the one or more programs include instructions for executing the source-grid-load-storage collaborative planning method that considers multi-agent hybrid games as described above.

[0028] Compared with the prior art, the present invention has the following beneficial effects:

[0029] 1. The present invention constructs a distribution network collaborative planning model that takes into account the master-slave-cooperative mixed game of source, grid, load and storage entities. It innovatively combines the master-slave game relationship between distribution network operators and source, grid, load and storage entities with the cooperative game within the source, grid, load and storage entities, effectively stimulating the enthusiasm of market entities to participate in planning, effectively balancing the interests of entities on all sides of the source, grid, load and storage, promoting the consumption of new energy, reducing the system's network supply load, improving the electricity consumption behavior of power users, realizing flexible and efficient configuration and coordinated utilization of source, grid, load and storage power resources, and improving the level of distribution network planning.

[0030] 2. The present invention balances the interests of multiple parties through a master-slave-cooperation hybrid game mechanism and optimizes the overall resource allocation efficiency. The upper master-slave game sub-model aims to maximize the comprehensive benefits of distribution network operators, and the lower cooperative game sub-model is used to stimulate new energy investment and user-side demand response, reduce dependence on the main grid, improve system flexibility and new energy absorption capacity, solve the problem of insufficient consideration of individual interests in traditional methods, and significantly improve the rationality of planning schemes.

[0031] 3. The present invention adopts a model solving method based on target cascade analysis method and adaptive alternating direction multiplier method to realize the decoupling and distributed iterative solution of the double-layer model, which can achieve global convergence, efficient solution and reliable solution results. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 It is a schematic diagram of the multi-agent mixed game relationship of the present invention;

[0033] Figure 2 This is a schematic diagram of the overall solution process of the distribution network collaborative planning model considering the master-slave-cooperative hybrid game of the source, grid, load and storage entities of the present invention;

[0034] Figure 3 Planning an initial topology for the IEEE 33 node system in an embodiment of the present invention;

[0035] Figure 4 The typical daily output data in the embodiment of the present invention;

[0036] Figure 5 This is a diagram showing the simulated transaction results of a distribution network operator in scenario 2 in an embodiment of the present invention;

[0037] Figure 6 It is a power result diagram of demand response electric load in an embodiment of the present invention;

[0038] Figure 7 This is a diagram showing the results of a simulated transaction for power users in scenario 2 in an embodiment of the present invention;

[0039] Figure 8 It is a diagram of the iterative convergence process of the algorithm in the embodiment of the present invention;

[0040] Fig. 9 Graph showing changes in the Q1 objective function of each subject in an embodiment of the present invention. DETAILED DESCRIPTION

[0041] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. This embodiment is implemented based on the technical solution of the present invention, and provides a detailed implementation method and specific operation process, but the protection scope of the present invention is not limited to the following embodiments.

[0042] The present embodiment provides a source-grid-load-storage collaborative planning method considering a multi-agent hybrid game, the method comprising the following steps: establishing a distribution network collaborative planning model considering a master-slave-cooperative hybrid game of source-grid-load-storage subjects, the source-grid-load-storage subjects comprising a distribution network operator, a distributed power source operator, an energy storage operator and a power user, the distributed power source operator, the energy storage operator and the power user forming a source-load-storage side subject, the distribution network collaborative planning model being a two-layer model, comprising an upper master-slave game sub-model between the distribution network operator and other subjects and a lower cooperative game sub-model between the source-load-storage side subjects; adopting a solution strategy combining a target cascade analysis method and an adaptive coefficient alternating direction multiplier method to solve the distribution network collaborative planning model, and obtaining the optimal planning scheme for new line construction, distributed power sources and energy storage devices.

[0043] The above method innovatively combines the master-slave game relationship between distribution network operators and source-grid-load-storage entities with the cooperative game within the source-grid-load-storage entities, and solves it based on the target cascade analysis method and the adaptive alternating direction multiplier method. It can balance the interests of multiple parties, optimize the overall allocation efficiency of resources, and improve the rationality and reliability of distribution network planning.

[0044] 1. Game Relationship

[0045] This embodiment involves a game relationship between four stakeholders and two planning levels. The distribution network operator decides on the new line construction plan with the goal of reducing line expansion costs, network losses and operating costs; the distributed generation (DG) operator and the energy storage operator decide on the capacity configuration and access location of the DG and energy storage devices with the goal of reducing investment costs and increasing electricity sales revenue respectively; the power user decides on its electricity consumption mode based on the electricity price signal and incentive mechanism with the goal of minimizing the electricity cost. The grid line will affect the site selection and capacity determination of DG and energy storage, and the access location capacity of DG and energy storage will affect the electricity sales revenue of the distribution network operator and the electricity purchase choice of the power user, while the power user's electricity consumption plan will affect the kilowatt-hour price and profit income of the distribution network operator, DG operator and energy storage operator. However, the interests of all parties are compatible with each other to a certain extent, and there is the possibility of win-win cooperation. DG and energy storage relieve the power supply pressure of the distribution network during operation, and the user-side demand response promotes the balance of power supply and demand and ensures the stable operation of the system. Direct market transactions between DG, energy storage and electricity users break the monopoly of distribution network operators, increase the selectivity of power interaction, and to a certain extent avoid the phenomenon of wind and solar power abandonment and increase the utilization rate of energy storage batteries.

[0046] refer to Figure 1As shown in the figure, the distribution network operator leads the construction and management of the distribution network and has a high decision-making power, and can be regarded as the leader of the game. In the upper-level problem, the distribution network operator establishes the grid structure of the line and passes it to other entities that follow in the lower level. The lower-level entities have a cooperative game relationship. In addition to purchasing and selling electricity from the distribution network, they can conduct market transactions independently. DG operators and energy storage operators determine the access capacity and location based on the existing topological structure and optimize the operation plan. At the same time, power users decide on the electricity consumption mode and purchase options based on the electricity price signal, and pass the power generation and consumption plan to the upper-level distribution network operator. The distribution network operator re-plans the grid and sets the electricity price based on the latest configuration plan and power generation and consumption plan. The two-level game coordinates and iterates until equilibrium is reached, achieving mutual benefit and win-win results for all four parties.

[0047] 2. Distribution Network Collaborative Planning Model

[0048] 1. Upper-level master-slave game sub-model

[0049] In this embodiment, the upper master-slave game sub-model is a planning decision-making model led by the grid-side subject, considering the construction of new grid lines, with the goal of maximizing the annual comprehensive profit of the distribution network operator, and the objective function is:

[0050]

[0051] Where: C DSO , They are the annual comprehensive income, annual electricity sales income, annual electricity purchase cost from the upper grid and annual investment cost of the distribution network operator; t T 、f t G are the electricity sales price and the electricity price of the upper power grid in period t respectively; They are the electricity sold in scenario s and the electricity purchased by the substation in period t; is the set of lines to be newly built; i is the length of the i-th newly built line; is the unit length investment cost of the i-th new line; A 0-1 variable indicating whether the selected line is newly built; N s is the number of scenes; D s is the number of days in scenario s; b is the discount rate; y is the useful life of the equipment.

[0052] The constraints considered in the upper master-slave game sub-model include system power flow constraints, system operation safety constraints, new line constraints and transaction power balance constraints.

[0053] 1) System power flow constraints

[0054]

[0055] Where: u(j) is the set of the head nodes of the branch with j as the terminal node; w(j) is the set of the terminal nodes of the branch with j as the head node; P ij,s,t , P jk,s,t and Q ij,s,t , Q jk,s,t are the active and reactive powers flowing through feeders (i, j) and (j, k) in scenario s and period t, respectively; U i,s,t , U j,s,t I is the voltage amplitude of nodes i and j in time period t of scenario s; ij,s,t is the current amplitude of line ij in scenario s and period t; P j,s,t and Q j,s,t are the active and reactive power injected into node j respectively; and is the active and reactive power of DG at node j; is the charging and discharging power of the energy storage at node j; and are the load interruption power, outward transfer power, and inward transfer power at node j, respectively; and is the active and reactive power of the load at node j; R ij , X ij , Z ij are the resistance, reactance and impedance of line ij respectively.

[0056] 2) System operation safety constraints

[0057] U min ≤U i,t ≤U max (10)

[0058]

[0059] Where: U max and U min is the voltage upper and lower limits; I ij,max is the safety current of line ij; is a 0-1 variable, indicating whether line ij is connected; U i,t ,I ij,t is the voltage amplitude of node i and the current amplitude of line ij during period t.

[0060] 3) New line constraints

[0061]

[0062] Where: is the set of lines to be built at the newly added node a; Indicates whether line i at the newly added node a is newly created.

[0063] 4) Transaction power balance constraints

[0064]

[0065] Where: are the transaction powers of DG operators, energy storage operators, power users and stakeholder m in scenario s and period t respectively; Φ0 is the set of other stakeholder entities except itself.

[0066] 2. Lower-level cooperative game sub-model

[0067] The lower-level cooperative game sub-model is constructed based on the planning and decision-making model of the source-load-storage following subject, which includes DG operators, energy storage operators and power users. The details are as follows:

[0068] DG operator: Considering the interests of DG operators mainly engaged in photovoltaic and wind power, the installation location and capacity of distributed wind power and distributed photovoltaic are determined with the goal of maximizing the annual comprehensive income of the operator. The objective function can be uniformly expressed as:

[0069]

[0070] Where: C DG , They are the annual comprehensive income of DG operators, annual electricity sales income, government new energy subsidies, annual investment costs and annual operation and maintenance costs; They are the electricity sales price of the DG operator, the government's new energy subsidy price, and the DG unit power generation operation and maintenance fee; Ω DG Alternative installation node set for DG; c DG is the DG unit capacity investment cost; Install capacity for DG at node i; It is a 0-1 variable, indicating whether the i-th candidate node is connected to the DG.

[0071] The constraints include:

[0072] 1) DG installation capacity constraints

[0073]

[0074] Where: is the maximum installed capacity of DG at node i.

[0075] 2) DG operation constraints

[0076]

[0077] Where: They are the upper and lower limits of the active and reactive output of DG at node i in scenario s and period t, respectively.

[0078] 3) DG transaction power balance constraints

[0079]

[0080] Where: is the transaction power between DG and stakeholder m at node i in scenario s and period t.

[0081] Energy storage operators: Energy storage operators use low storage and high sales to arbitrage electricity, with the goal of maximizing their annual comprehensive income, and determine the installation location and capacity of energy storage devices. The objective function is as follows:

[0082]

[0083] Where: C ESS , They are the energy storage operator’s annual comprehensive income, annual electricity sales income, new energy subsidy costs, annual investment costs, and annual operation and maintenance costs; They are the electricity sales price, electricity purchase price, new energy subsidy price, and operation and maintenance unit price of the energy storage operator when charging and discharging a unit of electricity; Ω ESS Alternative installation node sets for energy storage; is the investment price per unit capacity of energy storage; installing capacity for energy storage; is the investment decision variable for energy storage.

[0084] The constraints include:

[0085] 1) Energy storage installation capacity constraints

[0086]

[0087] Where: It is the maximum installed capacity of ESS.

[0088] 2) Energy storage operation constraints

[0089]

[0090] Where: Indicates the energy storage charging and discharging state at node i in scenario s and time period t. It is a 0-1 variable, 1 for discharging and 0 for charging. is the energy storage capacity at node i in scenario s and period t; are the energy storage charging and discharging efficiency respectively; They are the upper and lower limits of energy storage capacity respectively; are the energy storage capacity at the initial and final moments of operation respectively; Δt is the duration of time period t.

[0091] 3) Energy storage trading power balance constraints

[0092]

[0093] Where: is the transaction power between the energy storage at node i and the stakeholder m in scenario s and period t.

[0094] Power users: Considering two types of demand response loads, interruptible loads and transferable loads, power users adjust their power consumption behavior based on power price signals and grid incentive mechanisms to maximize their own power cost reduction. The objective function is as follows:

[0095] max C DR =C IL +C TL (35)

[0096]

[0097] Where: C DR , C IL , C TL They are the electricity cost reduction after the power users participate in demand response, the subsidy income of interruptible load, and the electricity cost reduction of transferable load; f IL 、f TL It is the interruptible load compensation price and the load purchase price; are the load interruption power, outward transfer power, and inward transfer power at node i in scenario s and period t.

[0098] 1) Demand response load constraints

[0099]

[0100] Where: and and and They are the upper and lower limits of the interruptible load power and the upper and lower limits of the transferred in / out load power at node i in scenario s and period t, respectively.

[0101] 2) Power balance constraints for power users

[0102]

[0103] Where: is the transaction power between the power user and stakeholder m at node i in scenario s and period t.

[0104] Furthermore, the distribution network collaborative planning model of the leader and the follower in the master-slave game stage in this embodiment is as follows:

[0105]

[0106] Where: C UThe annual comprehensive benefits of the alliance.

[0107] Followers form a stakeholder alliance and respond to the decisions of distribution network operators through cooperation. The goal of cooperation is to maximize the overall benefits, while ensuring that their own benefits will not be damaged after cooperation. In cooperative games, the Nash negotiation method can effectively solve the game between multiple stakeholders and effectively describe the cooperative interaction between stakeholders on the source, load and storage side of the market. Its standard model is:

[0108]

[0109] Where: F n is the benefit of subject n after cooperation; is the income of subject n before cooperation, that is, the negotiation breakdown point; F is the number of subjects.

[0110] The above standard model is a non-convex nonlinear problem of multi-variable coupling, which can be decomposed into two sub-problems and solved in sequence. The equilibrium solution of the model represents that the benefits of the negotiating entities have reached Pareto optimality.

[0111] Furthermore, in this embodiment, the lower-level cooperative game sub-models are divided into a comprehensive cost minimization sub-problem and a benefit distribution maximization sub-problem.

[0112] The comprehensive cost minimization sub-problem (Q1) is as follows:

[0113] The comprehensive cost is the sum of the electricity costs of DG operators, energy storage operators, and power users. The objective function is:

[0114]

[0115] Where: C n is the annual comprehensive cost of entity n.

[0116] The constraints are as follows:

[0117] 1) Transaction power constraints

[0118] |P n-m,s,t |≤P n-m,s,max (47)

[0119] Where: P n-m,s,t , P n-m,s,max is the transaction power and transaction power upper limit between stakeholder n and stakeholder m.

[0120] 2) Electricity trading balance constraints

[0121]

[0122] Where: P n,s,t is the total output power of stakeholder n in scenario s and period t.

[0123] The sub-problem of maximizing the benefit distribution (Q2) is as follows:

[0124] After achieving the minimum comprehensive cost, consider how to distribute the benefits according to the power contribution of each entity in the cooperation. The greater the power contribution, the greater the bargaining power of the entity in the distribution of benefits. It is generally believed that providing power contributes more than consuming power. An power mapping function is established based on the exponential function of natural constants to measure the bargaining power:

[0125]

[0126]

[0127] Where: are the total energy provided and consumed by stakeholder n respectively; are the maximum energy provided and consumed by the alliance entity respectively; α n is the bargaining power of the interest entity n.

[0128] The sub-problem of maximizing the distribution of benefits aims to maximize the incremental benefits of the subject after the cooperation compared with before the cooperation, and is converted into a minimum value problem in the form of a logarithmic function of a strictly monotonically increasing convex function:

[0129]

[0130] Where: The benefits after cooperation of the interest subject n; is the benefit of the interest subject n before cooperation; w n-m,s,t is the transaction electricity price between stakeholder n and stakeholder m in scenario s and period t.

[0131] The constraints are as follows:

[0132] 1) Cooperation benefit constraints

[0133] Every entity that participates in the cooperation and makes a contribution can gain benefits.

[0134]

[0135] 2) Trading electricity price constraints

[0136] w s,t,min ≤w n-m,s,t ≤w s,t,max (55)

[0137] Where: w s,t,max 、w s,t,min They are the upper and lower limits of the trading electricity price in scenario s during period t.

[0138] 3. Solution Strategy

[0139] In this embodiment, when solving the distribution network collaborative planning model, the target cascading analysis method (ATC) is first used to decouple the two-layer model, and then the adaptive coefficient alternating direction multiplier method (ADMM) is used to distribute the lower cooperative game sub-model, decomposing the lower sub-problem into multiple independent optimization models, and each subject can solve autonomously and achieve global convergence through alternating iterations. Figure 2 shown.

[0140] Since the interaction between the distribution network operator and other stakeholders' alliance exists in the upper model and Q1, Q2 determines the transaction price only based on the solution value of the transaction volume between the alliance entities. During the interaction process, the solution of Q2 will not affect the solution of Q1, nor will it be passed to the upper problem. Therefore, the coupling of the two-layer model only exists in the upper problem and Q1, satisfying the following consistency constraints:

[0141]

[0142] Where: The power that the stakeholder alliance expects to trade with the distribution network operator; The power that the distribution network operator expects to trade.

[0143] The constraint is decoupled and the corresponding penalty term is added to the objective function to relax the consistency constraint. The decoupled objective is:

[0144]

[0145] Where: k, k o is the number of inner and outer loop iterations of the target cascade analysis method.

[0146] The outer loop converges when:

[0147]

[0148] To ensure the convergence of the model, the penalty factor can be updated at the end of the external iteration:

[0149]

[0150] In the formula: 1<σ<3 is usually taken.

[0151] After decoupling, distribution network operators optimize the grid structure in the upper-level problem, and the alliance of stakeholders optimizes the configuration plan and electricity usage method in the lower-level problem. The upper and lower-level models interactively access power information to guide the planning decisions of each entity.

[0152] Furthermore, in the two sub-problems of the lower model, there are multiple couplings between the power trading balance and the power price balance, and auxiliary variables can be introduced to convert the original constraints into double-coupled constraints. The establishment of the equation indicates that cooperation has been reached between the two entities.

[0153]

[0154] Where: P m-n,s,t is the transaction power between interest subject m and interest subject n; w m-n,s,t is the transaction electricity price between interest subject j and interest subject i.

[0155] After decoupling, the alternating direction multiplier method can be used to distribute the two sub-problems. First, solve Q1 to obtain the transaction power between each entity, and then solve Q2 to obtain the expected transaction electricity price of the entity. The specific steps are as follows:

[0156] 1) Establish the augmented Lagrangian function of the subproblem.

[0157]

[0158] Where: is the Lagrange multiplier; is the penalty factor.

[0159] 2) Based on the expected transaction volume / price information provided by other entities, each entity independently updates its decision and transmits its expected transaction volume / price to other entities. This process is repeated until each cooperative entity has updated its decision.

[0160]

[0161] Where: k is the number of iterations.

[0162] 3) Update the Lagrange multiplier.

[0163]

[0164] 4) Adaptive penalty coefficient adjustment.

[0165] The value of the penalty coefficient will affect the convergence speed and the feasibility of the solution. In order to improve the convergence speed of the alternating direction multiplier method and make the original feasibility and the dual feasibility decrease at a consistent rate, the penalty coefficient can be dynamically adjusted in each iteration. The present invention introduces an adaptive adjustment mechanism to dynamically correct the penalty coefficient:

[0166]

[0167] Where: is the original residual and dual residual of Q1 at the kth iteration; δ>1,τ1>1,τ2>1, usually δ=10,τ1=τ2=2. Similarly, it can be adjusted dynamically

[0168] 5) If the conditions are met, the kth iteration converges and the iteration terminates.

[0169]

[0170] 6) If the number of iterations k>k max , the algorithm does not converge and the iteration terminates; if it is not satisfied, set k=k+1 and return to step 2) to continue iterating and solving.

[0171] If the above method is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0172] In order to verify the effectiveness of the source-grid-load-storage planning method proposed in the present invention, this embodiment is set up based on the IEEE 33-node system. Due to the growing load demand, the distribution network adds load nodes 34 to 37, and their load data are shown in Table 1. After the increase, the maximum load of the system is 4175kW+j2560kvar. The power purchase price of the distribution network operator to the superior power grid is 0.38 yuan / (kW·h). Figure 3 The red dotted line in the middle shows the length of the alternative line and other data as shown in Table 2. The cost of the new line is 100,000 yuan / km, and the resistance and reactance are 0.25Ω / km and 0.4Ω / km. The planning period is 10 years, and the discount rate is 6%.

[0173] The relevant parameters of DG and energy storage are shown in Table 3. Assuming that the user side is all transferable load, participating in price-based demand response, the time-of-use electricity price is shown in Table 4. The 25 nodes of the system are used as interruptible load nodes, the interruptible time is 11:00-22:00, and the interruptible load subsidy is 0.4 yuan / (kW·h). Based on the historical data of the distribution network of a certain place throughout the year, the typical daily scenario is obtained by Gaussian mixture clustering. Figure 4 shown.

[0174] Table 1 New node load data

[0175]

[0176] Table 2 Alternative line parameters

[0177]

[0178] Table 3 Related parameters

[0179]

[0180]

[0181] Table 4 Electricity price parameters

[0182]

[0183] In order to verify the effectiveness of the distribution network source-grid-load-storage collaborative planning method based on hybrid game proposed in this invention, the following three methods are used to solve the problem and compare the planning results.

[0184] Method 1: The method of the present invention is a two-level programming of multi-agent hybrid game considering master-slave cooperation.

[0185] Method 2: Distribution network planning method based on master-slave game. That is, only the master-slave game between the distribution network operator and other entities is considered. At this time, the planning goal is to maximize the comprehensive benefits of the distribution network operator and the source-load-storage entity alliance.

[0186] Method 3: Traditional distribution network collaborative planning method. That is, the distribution network source, grid, load and storage planning without considering the game relationship between multiple subjects. At this time, the planning goal is to maximize the overall planning benefit.

[0187] Methods 1, 2, and 3 compare and analyze the necessity of considering multi-agent game in distribution network planning. Methods 1 and 2 compare and analyze the results of stakeholder participation in planning under different game relationships, and verify the effectiveness of follower cooperation in the source-load-storage side game.

[0188] The planning results of different methods are shown in Table 5. In Method 1, the DG operator has the largest access capacity for photovoltaic and wind turbines. This is because when considering the interests of the subject, if the wind and solar power can be consumed locally, the DG operator will increase the investment in DG to increase the operating income. In Method 3, which aims to maximize the overall benefit, the distribution network operator has a lower purchase fee from the upper power grid, so the demand for DG is relatively small. From the line planning results, it can be seen that the length of each newly built line in Method 3 is the shortest, because for the planning goal of maximizing the overall benefit, the smaller the line impedance, the smaller the loss, and the greater the investment and operating income while ensuring the safety of the system and giving full play to the flexibility of resources. In Methods 1 and 2, there is a game between the distribution network operator and other subjects. During planning, the power generation and consumption decisions of each subject will affect the distribution network's selection of new lines to a certain extent. For example, in Method 1, node 21 has distributed wind power access. For DG operators, building a new line at node 21 or 22 to connect node 35 can expand the radial network, which is conducive to the transaction between DG operators and power users. At the same time, line 22-35 is shorter than line 21-35, so the final new line plan is 22-35. From the planning results of energy storage, it can be seen that method 1 has the largest energy storage access capacity. In method 1, energy storage operators can cooperate with DG operators and electricity users to purchase energy storage at a lower electricity price, and give priority to selling it to nearby loads when the electricity price is higher, so as to promote the utilization of energy storage devices and increase profits. From the above analysis, it can be seen that, considering the influence of multi-agent game, the grid-connected capacity of distributed energy after planning increases, which helps to reduce the dependence of distribution networks on the upper power grid, promote energy utilization, and reduce carbon emissions.

[0189] Table 5 Planning results of different methods

[0190]

[0191] The annual comprehensive income / expenses of each stakeholder planning obtained by different methods are shown in Table 6. It can be seen that the income / expenses of each operator in Method 2 are higher than those in Method 3. The planning income of DG operators and distribution network operators in Method 1 is relatively small. This is because the stakeholders have bargaining power. During planning, the operators can propose price demands based on the expected amount of electricity to be interacted. The electricity price when reaching a consensus is often lower than the initial price to promote the interaction of electricity between the two parties. However, overall, the annual comprehensive total cost of the main planning of the source-grid-load-storage side in Method 1 is 5.1111 million yuan, which is 94,700 yuan and 266,700 yuan less than that of Method 2 and Method 3, respectively. Therefore, considering multiple subjects in planning can maximize the overall benefits by transferring the interests of some subjects, increase the enthusiasm of other subjects to participate in the construction of the distribution network, and promote the rational allocation of resources.

[0192] Table 6 Benefits / costs of stakeholder planning

[0193]

[0194] Combined with the planning results, the economic efficiency of each subject in the planning under the mixed game mode is analyzed by comparing the various expenses of the stakeholders under different methods.

[0195] The various benefits and costs of distribution network operators under different planning methods are shown in Table 7. It can be seen that in the planning considering the mixed game, the distribution network purchases less electricity from the upper power grid, the electricity sales revenue decreases, and the investment cost increases slightly. The simulated transaction results of the distribution network operator in the typical day scenario 2 are shown in Figure 5 As shown. It can be seen that the distribution network mainly obtains revenue by selling electricity to power users, and has less interaction with DG and energy storage, so the distribution network's revenue from electricity sales is less than other methods. The large-scale access of distributed energy in Method 1 has alleviated the problem of increased network losses in the distribution network caused by the influence of cooperative entities on line planning to a certain extent. Therefore, the cooperation of entities on the source, load and storage side will reduce the planning revenue of the distribution network operator, but from the perspective of the overall system, it can reduce the distribution network's dependence on the upper power grid and improve power supply flexibility.

[0196] Table 7 Revenues and expenses of distribution network operators

[0197]

[0198] The various benefits and costs of DG operators are shown in Table 8. In method 1, DG has the largest access capacity, so the investment cost is higher. In the operation transaction, the transaction power of method 1 is the largest. Due to the influence of the cooperative game with other alliance entities, the DG power selling price is lower, so the power selling income is lower than that of method 2.

[0199] Table 8 DG operator's revenue and expenses

[0200]

[0201] The various benefits and expenses of energy storage operators are shown in Table 9. In Method 1, the energy storage capacity configuration increases, and the energy storage charging and discharging volume increases, so that transactions with DG operators and arbitrage of low storage and high generation strategies can be effectively utilized to increase operating income. Therefore, through cooperation with the stakeholders on the source and load side, the investment capacity of energy storage devices can be increased, the comprehensive benefits of energy storage operators can be improved, and the stakeholders on the storage side can be actively involved in planning.

[0202] Table 9 Revenues and expenses of energy storage operators

[0203]

[0204] All power loads are set to participate in price-based demand response, and the various fees for power users are shown in Table 10. In the three methods, the interruptible loads all actively participate in the response and make decisions on the maximum interruptible power to reduce the power purchase cost and obtain interruptible subsidies.

[0205] Table 10 Various fees for electricity users

[0206]

[0207] Taking typical day scenario 2 as an example, the demand response load power and simulated transaction results of power users in method 1 are as follows: Figure 6 and Figure 7 As shown. 11:00-13:00 and 18:00-22:00 are peak electricity price periods, during which electricity users transfer their loads to off-peak periods and normal periods to reduce electricity expenses. At the same time, during the period 11:00-22:00, electricity users interrupt part of their loads according to their own electricity needs. From the results in Table 10, it can be seen that the electricity costs of electricity users in method 1 are reduced the most, which is 3.2405 million yuan less than that in method 2. This is because in the cooperative game, electricity users can give priority to trading with other entities except distribution network operators, and participate in negotiation and bargaining to determine the electricity purchase price. Figure 7 As shown in the figure, during the period when DG power generation is low, power users have greater bargaining power, so they can buy electricity at a price lower than the original electricity price of the DG operator. During the period of 8:00-19:00, the photovoltaic power generation increases, and the bargaining power of the DG operator is stronger at this time, and the photovoltaic power selling price increases. When the DG power generation cannot meet the demand, power users will purchase part of the electricity from the energy storage operator at a price lower than the time-of-use electricity price. Based on this, through the cooperation of the interest group alliance, the economic benefits of the load side in the plan can be effectively improved, and the enthusiasm of power users can be enhanced.

[0208] The convergence of the algorithm used in the present invention is as follows Figure 8 As shown in the figure, in the outer loop of the target cascade analysis algorithm, convergence is achieved after 8 iterations. In the last outer loop, the dual residual of subproblem Q1 converges to within 10-1 after 25 iterations through the adaptive coefficient alternating direction multiplier method. At this time, under the decision of line planning and site selection and capacity determination, the stakeholders reach a consensus on power trading. In the benefit distribution stage, subproblem Q2 successfully converges after 10 iterations.

[0209] In order to verify the effectiveness of dynamically updating the penalty coefficient through the adaptive adjustment mechanism in the process of solving the alternating direction multiplier method, taking subproblem Q1 as an example, different initial penalty coefficients are set for comparison. Considering that if the penalty coefficient is not selected appropriately, the solution time of the classic alternating direction multiplier method is too long, according to ρ = 10 2The adaptive coefficient alternating direction multiplier method converges after 25 iterations, and the maximum number of iterations is set to 25. Analysis ρ = 10 2 and ρ = 10 4 The effects of the two solutions on the changes in the objective function values ​​of each subject under the two values ​​are shown in the following table. Fig. 9 As shown in the figure, under the classical alternating direction multiplier method, ρ = 10 2 When ρ = 10, the objective function does not reach the optimal value after 25 iterations and converges slowly, indicating that the penalty coefficient is too small at this time; 4 When , the objective function value shows an oscillating characteristic in the early iteration process, indicating that the penalty coefficient is too large at this time. For the adaptive coefficient alternating direction multiplier method, the initial value of ρ is 10 2 When , the convergence speed of the objective function value of each subject is faster than that of the classical alternating direction multiplier method; the initial value of ρ is 10 4 , the oscillation process in the iteration is weakened. This is because, if ρ is small, dynamically increasing ρ can increase the weight of the penalty term of the objective function and accelerate the convergence process; if ρ is large, dynamically reducing ρ can avoid the Lagrange multiplier from growing too fast, thereby weakening the oscillation. Therefore, by adaptively adjusting the penalty coefficient, the initial value of ρ can be set within a larger range, effectively avoiding the difficulty in solving caused by improper selection of the penalty coefficient. In addition, in the two-level programming model established by the present invention, after reaching the convergence condition of Q1, the convergence requirement of the external loop must also be met. Therefore, dynamically adjusting the penalty coefficient during the alternating iteration process can reduce the number of internal loops in each external loop process and improve computational efficiency.

[0210] In other embodiments, a source-grid-load-storage collaborative planning system considering multi-agent hybrid games may also be provided, comprising one or more processors, a memory, and one or more programs stored in the memory, wherein the one or more programs include instructions for executing the source-grid-load-storage collaborative planning method considering multi-agent hybrid games as described above.

[0211] The preferred specific embodiments of the present invention are described in detail above. It should be understood that a person skilled in the art can make many modifications and changes based on the concept of the present invention without creative work. Therefore, any technical solution that can be obtained by a person skilled in the art through logical analysis, reasoning or limited experiments based on the concept of the present invention on the basis of the prior art should be within the scope of protection determined by the claims.

Claims

1. A source-grid-load-storage collaborative planning method considering multi-agent hybrid game, characterized in that: The following steps are involved: A distribution network collaborative planning model considering the master-slave-cooperative hybrid game of source-grid-load-storage subjects is established. The source-grid-load-storage subjects include distribution network operators, distributed power generation operators, energy storage operators and power users. The distributed power generation operators, energy storage operators and power users constitute the source-load-storage side subjects. The distribution network collaborative planning model is a two-layer model, including an upper master-slave game sub-model between the distribution network operator and other subjects and a lower cooperative game sub-model between the source-load-storage side subjects. The distribution network collaborative planning model is solved by combining the target cascade analysis method with the adaptive coefficient alternating direction multiplier method to obtain the optimal planning scheme for new line construction, distributed power sources and energy storage devices.

2. The source-grid-load-storage collaborative planning method considering multi-agent hybrid game according to claim 1 is characterized in that: The upper-level master-slave game sub-model is a planning and decision-making model dominated by distribution network operators, with the goal of maximizing the annual comprehensive benefits of distribution network operators. The constraints considered include system flow constraints, system operation safety constraints, new line constraints and transaction power balance constraints.

3. The source-grid-load-storage collaborative planning method considering multi-agent hybrid game according to claim 1 is characterized in that: The lower-level cooperative game sub-model includes a comprehensive cost minimization sub-problem and a benefit distribution maximization sub-problem. The comprehensive cost minimization sub-problem aims to minimize the sum of the electricity costs of distributed power generation operators, energy storage operators, and power users, and the considered constraints include trading power constraints and electricity trading balance constraints. The benefit distribution maximization sub-problem aims to maximize the incremental benefits of each subject after participating in the cooperation compared to before the cooperation, and the considered constraints include cooperation benefit constraints and trading electricity price constraints.

4. The source-grid-load-storage collaborative planning method considering multi-agent hybrid game according to claim 3 is characterized in that: The objective function of the benefit allocation maximization sub-problem is expressed as: In the formula, The benefits after cooperation of the interest subject n; is the benefit of the interest subject n before cooperation; C n is the annual comprehensive cost of entity n; F is the number of entities; α n is the bargaining power of the interest subject n, which is obtained based on the electric energy mapping function.

5. The source-grid-load-storage collaborative planning method considering multi-agent hybrid game according to claim 3 is characterized in that: When solving the distribution network collaborative planning model, the target cascade analysis method is first used to decouple the two-layer model, and then the adaptive coefficient alternating direction multiplier method is used to perform a distributed solution to the lower-layer cooperative game sub-model.

6. The source-grid-load-storage collaborative planning method considering multi-agent hybrid game according to claim 1 or 5, characterized in that: In the target cascade analysis method, the end condition of the outer loop iteration is: In the formula, The power that the stakeholder alliance expects to trade with the distribution network operator; is the power that the distribution network operator expects to trade, with superscripts k and k o is the number of iterations of the internal and external loops of the target cascade analysis method, N s is the number of scenes, T is the number of time periods, ε o is the convergence threshold.

7. The source-grid-load-storage collaborative planning method considering multi-agent hybrid game according to claim 1 or 5, characterized in that: In the target cascade analysis method, the penalty factor is updated at the end of the outer loop iteration, and the update formula is: Where σ is the update constant, The kth o , k o +1 penalty factor for outer loop iteration.

8. The source-grid-load-storage collaborative planning method considering multi-agent hybrid game according to claim 5 is characterized in that: The distributed solution includes: first solving the comprehensive cost minimization sub-problem to obtain the transaction power between the entities, and then solving the benefit distribution maximization sub-problem to obtain the expected transaction electricity price of the entity.

9. The source-grid-load-storage collaborative planning method considering multi-agent hybrid game according to claim 8 is characterized in that: The distributed solution specifically includes: Establish the augmented Lagrangian function of each sub-problem; Based on the expected transaction volume / price information provided by other entities, each entity independently updates its decision and transmits the updated expected transaction volume / price information to other entities; Update the Lagrange multiplier and adaptively adjust the penalty factor; After the iteration termination condition is met, the final solution is obtained.

10. A source-grid-load-storage collaborative planning system considering multi-agent hybrid game, characterized in that: It includes one or more processors, a memory and one or more programs stored in the memory, and the one or more programs include instructions for executing the source-grid-load-storage collaborative planning method considering multi-agent hybrid game as described in any one of claims 1-9.

Images