Source network load storage multi-agent optimization scheduling method based on hybrid game theory

Abstract

The invention relates to the technical field of integrated energy system optimization scheduling, and discloses a source network load storage multi-agent optimization scheduling method based on a hybrid game theory. Firstly, a source network load storage framework considering electric energy transaction is constructed, a power distribution network operator and integrated energy system cooperative alliance adopts a master-slave game, and different integrated energy systems perform energy interaction in a cooperative game mode; comprehensively considering multi-party interest demands of a distributed power supply operator, a power distribution network operator, an energy storage operator and a demand side, and constructing a source network load storage multi-subject optimization scheduling model considering multi-subject interests; a particle swarm algorithm is combined with a deep learning algorithm based on ADMM decomposition, the particle swarm algorithm solves a master-slave game, the deep learning algorithm cooperates with the game, and the particle swarm algorithm and the deep learning algorithm are combined to solve the constructed model. Through the construction of the source-load-storage cooperative alliance, the operation cost of each operation main body is remarkably reduced, the consumption of renewable energy sources is promoted, and reasonable distribution of cooperative profits is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of optimized scheduling of integrated energy systems, and in particular to a source-grid-load-storage multi-agent optimized scheduling method based on hybrid game theory. Background Art

[0002] To achieve the goals of "carbon peak and carbon neutrality", building a new power system with integrated operation of source, grid, load and storage is an effective method, but the increase in the proportion of renewable energy power generation brings challenges to the operation of the power grid. Among them, the integrated energy system (IES) promotes the consumption of renewable energy through resource integration. With the development of IES, there are multiple IES operating in the same distribution network area, which can form an IES cooperative alliance. The IES cooperative alliance fully utilizes the idle resources of each IES through peer-to-peer (P2P) electricity transactions between members of the cooperative alliance, and improves the demand response capability of the IES cooperative alliance.

[0003] However, IES alliance members represent diverse stakeholders, making traditional single-agent optimization scheduling methods inapplicable. Therefore, designing an effective scheduling strategy for inter-IES energy transactions is crucial. Existing research often uses game models to address the complex inter-IES interests. Literature has documented the use of a master-slave game to manage inter-microgrid energy transactions, establishing a multi-IES optimization scheduling model with a multi-microgrid system as the leader and a load aggregator as the follower. Alternatively, a master-slave game-based optimization configuration model for a multi-microgrid system is established with a multi-microgrid operator as the leader and the distribution network as the follower. However, in a master-slave game, both the leader and the follower aim to maximize their own interests, failing to achieve optimal overall performance. However, managing energy interactions through cooperative games can balance individual and overall interests. Examples include multi-microgrid cooperative game models described in existing literature and the cooperative game Nash negotiation model for IES alliances based on Nash negotiation theory.

[0004] While the aforementioned research considers multi-microgrid optimal scheduling through inter-microgrid energy trading, it still suffers from the following shortcomings. First, by constructing a cooperative game model, the IES alliance passively accepts the decisions of the distribution network, preventing effective interaction between the distribution network and the IES alliance. Second, the IES alliance exchanges energy with the grid at a fixed price, which makes the IES alliance dependent on the grid for load balancing, thus hindering alliance members' enthusiasm for participating in demand response and inter-IES energy exchange.

[0005] In summary, existing research on the coordinated operation and scheduling of sources, grids, loads and storage mainly focuses on the analysis of diversified and differentiated demands under the coordinated optimization of various entities. There is still a lack of research on designing effective electricity and electricity price trading mechanisms to meet the needs of multiple energy sources and multiple entities. Summary of the Invention

[0006] To address the above issues, the present invention aims to provide a multi-agent optimization scheduling method for power generation, grid generation, load generation, and storage based on hybrid game theory. This method proposes a hybrid game optimization model between distribution network operators and power generation, load generation, and storage. This model manages the competitive or cooperative relationships among different stakeholders through a master-slave game nested within a cooperative game, enabling the dynamic formulation of electricity trading prices and interactive optimization of electricity across integrated energy systems. The technical solution is as follows:

[0007] A multi-agent optimization scheduling method for source, grid, load and storage based on hybrid game theory includes the following steps:

[0008] Step 1: Construct a hybrid game model of the source-grid-load-storage framework considering power trading:

[0009] The distribution network operators and the integrated energy system cooperative alliance adopt a master-slave game. The distribution network operators are the leaders, who influence the dispatching strategy of the integrated energy system by setting energy transaction prices to maximize profits. The integrated energy system is the follower, who accepts the prices set by the distribution network and changes its energy consumption pattern to reduce its own operating costs.

[0010] At the same time, different integrated energy systems use cooperative game theory to exchange energy. When there are differences in renewable energy generation and load demand at different times among integrated energy systems, the integrated energy systems achieve energy sharing and fair distribution of benefits through cooperation. Each integrated energy system uses Nash negotiation to determine the electricity transaction volume and transaction price.

[0011] Step 2: Construct a multi-agent optimization scheduling model for source, grid, load and storage:

[0012] Based on the source-grid-load-storage framework, a multi-agent optimization scheduling model for source-grid-load-storage is constructed, taking into account the interests of distributed power generation operators, distribution network operators, energy storage operators, and the demand side. This model includes a distribution network operation model, a distributed power generation operator operation model, an energy storage operator operation model, and a demand response operation model.

[0013] Step 3: Solve the hybrid game of source, grid, load and storage

[0014] The particle swarm optimization algorithm is combined with a deep learning algorithm based on alternating direction multiplier decomposition. The particle swarm optimization algorithm solves the master-slave game, and the deep learning algorithm based on alternating direction multiplier decomposition solves the cooperative game. The two jointly solve the constructed hybrid game model and obtain the multi-agent optimization scheduling parameters of the source, grid, load and storage. Specifically:

[0015] First, the population of electric energy trading volume and the number of iterations of speed are initialized. Then the displacement and speed of each particle are updated and the fitness is calculated. If the difference between the previous electric energy trading volume and the current electric energy trading volume is within the set threshold and convergence is reached, the master-slave game is solved and ended. Otherwise, the historical optimal displacement of each particle and the historical optimal displacement of all particles are updated and iterations continue. After the master-slave game is solved, the objective functions of different integrated energy systems are iterated again using alternating direction multiplier decomposition, and the deep reinforcement learning method is used to solve and obtain the P2P energy trading price.

[0016] The beneficial effects of the present invention are:

[0017] 1) This paper proposes a hybrid game framework for the cooperative alliance of distribution network and integrated energy system, establishes a master-slave game model based on Stackelberg theory, and constructs a cooperative game model among members of the cooperative alliance of integrated energy system through Nash negotiation theory.

[0018] 2) The present invention adopts the particle swarm optimization algorithm combined with the deep learning algorithm based on ADMM decomposition to solve the constructed hybrid game model, verifying the rationality of the proposed method.

[0019] 3) The hybrid game model constructed in this invention can determine the purchase and sale prices of electricity within the integrated energy system cooperative alliance, as well as the electricity transaction prices between the system cooperative alliances. This model ensures the coordinated operation of the distribution network and the IES cooperative alliance while also enabling cooperation and profit distribution within the IES cooperative alliance. This model effectively improves the efficiency of each IES. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 A framework for electricity trading.

[0021] Figure 2 It is a mixed game negotiation framework.

[0022] Figure 3 It is a framework for interaction between multiple integrated energy systems and distribution networks.

[0023] Figure 4 The on-grid and off-grid electricity prices before and after the cooperation.

[0024] Figure 5 This is the optimization result of the interactive price of electricity.

[0025] Figure 6 This is the operating mode of the integrated energy park 1. DETAILED DESCRIPTION

[0026] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0027] The present invention introduces a distributed system operator (DSO) as the coordinator and operator of the IES cooperative alliance, and guides the operation optimization of the IES cooperative alliance by formulating electricity prices, so as to give full play to the scheduling potential of the IES cooperative alliance and reduce the IES cooperative alliance's dependence on the power grid. Due to the introduction of the DSO, the model needs to simultaneously optimize the electricity purchase and sales prices of the IES cooperative alliance to the DSO and the electricity exchange prices between the members of the IES cooperative alliance. To this end, the present invention proposes a hybrid game optimization model of DSO and IES cooperative alliance, in which the cooperative game of the IES cooperative alliance members is nested in the master-slave game between DSO and IES cooperative alliance to simultaneously deal with the competition and cooperation relationships in the model. In order to improve the solution efficiency of the model, the present invention combines the particle swarm algorithm with the alternating direction multiplier method (ADMM), wherein the particle swarm algorithm solves the master-slave game and the ADMM solves the cooperative game, and the two jointly solve the constructed model.

[0028] This paper uses the DSO's dynamic electricity pricing to coordinate the management of IES alliance unit scheduling, demand response, and peer-to-peer energy trading among members. A master-slave game model between the DSO and the IES alliance is established, with the cooperative game of IES alliance members embedded within it. Nash negotiation theory is used to allocate the alliance's cooperative benefits, which is then divided into two sub-problems: maximizing the alliance's benefits and allocating them. A particle swarm optimization algorithm (PSO) is proposed to solve the constructed model, and simulations are conducted to verify the feasibility and effectiveness of the method.

[0029] 1. Source-grid-load-storage framework considering power trading

[0030] The framework for electricity trading is as follows Figure 1 As shown in the figure, the distribution grid, as a leader, influences the scheduling strategy of the integrated energy system by setting energy transaction prices to maximize profits. As a follower, the integrated energy system accepts the prices set by the distribution grid and changes its energy consumption to reduce its own operating costs. Furthermore, different integrated energy systems use a cooperative game to exchange energy, reducing their dependence on the distribution grid.

[0031] First, the distribution network and the integrated energy system cooperative alliance engage in a master-slave game, considering the operating profits of the distribution network, the leader, and the multi-integrated energy system cooperative alliance. When energy exchange between integrated energy systems fails to meet their own needs or surplus electricity needs to be sold, the distribution network sets a real-time electricity transaction price based on the demand for electricity. Simultaneously, the integrated energy system accepts the electricity transaction price set by the leader and adjusts its electricity consumption.

[0032] Secondly, when there are differences in renewable energy generation and load demand across integrated energy systems at different times, integrated energy systems collaborate to achieve energy sharing and fair distribution of benefits. Each integrated energy system uses Nash negotiation to determine the electricity transaction volume and transaction price.

[0033] The process of the mixed game is as follows:

[0034] Step 1: According to the Nash-Harsanyi bargaining theory, the cooperative game problem of the integrated energy system cooperative alliance is divided into two sub-problems, namely, maximizing the interests of the cooperative alliance (P1) and allocating cooperative interests (P2).

[0035] Step 2: In a Stackelberg game (master-slave game), the distribution network operator acts as the leader, and the integrated energy system cooperative alliance acts as the follower. The distribution network operator aims to maximize its own profits, while the integrated energy system cooperative alliance aims to maximize the cooperative alliance's profits (P1). The electricity trading volume and price between the cooperative alliance and the distribution network operator, as well as the P2P energy trading volume and output value of each device within the integrated energy system, are determined.

[0036] Step 3: Based on the Nash-Harsanyi bargaining theory and the P2P contributions of each integrated energy system, achieve fair distribution of cooperative benefits (P2), and calculate the P2P energy transaction volume (electricity volume) and transaction price (electricity price).

[0037] 2. Multi-agent optimization scheduling model for power generation, grid, load and storage

[0038] In an open electricity market environment, the scope of distribution network stakeholders has gradually expanded. In addition to distribution network operators and electricity users, emerging stakeholders such as distributed power generation operators and energy storage operators have poured in.

[0039] The construction of distribution networks, the capacity allocation of distributed generation and energy storage, and user demand-side response are decided by multiple stakeholders. Different stakeholders have their own interests and demands for different planning objectives.

[0040] Taking into account the interests of multiple stakeholders during the distribution network planning phase helps stimulate their active participation, thereby improving overall planning. This multi-stakeholder collaborative planning model for the distribution network, combining source, grid, load, and storage, is constructed by comprehensively considering the interests of distributed generation operators, distribution network operators, energy storage operators, and the demand side.

[0041] (1) Distribution network operation model

[0042] Distribution network operators comprehensively consider the revenue from electricity purchase and sales, grid construction costs, and failure costs to ensure the safe and economical operation of the distribution network. The objective function is:

[0043] maxπ G =π G2IES +π G2PN -π O&M (1)

[0044] Where: π G represents the profit of the distribution network; π G2IES represents the transaction income between the distribution network and the integrated energy system; π G2PN Represents the transaction income between the distribution network and the main grid; π O&M Represents the line construction cost of the distribution network.

[0045]

[0046] Where: represents the price of electricity sold by the distribution network to the integrated energy system during the t-th operating period; represents the price of electricity purchased by the distribution network from the integrated energy system during the t-th operating period; represents the amount of electricity sold by the distribution network to the integrated energy system during the t-th operating period; It represents the amount of electricity purchased from the integrated energy system by the distribution network during the t-th operating period.

[0047]

[0048] Where: represents the price of electricity sold by the distribution network to the main grid during the t-th operating period; represents the price of electricity purchased by the distribution network from the main grid during the t-th operating period; represents the amount of electricity sold by the distribution network to the main grid during the t-th operating period; It represents the amount of electricity purchased by the distribution network from the main grid during the t-th operating period.

[0049]

[0050] Where: N l Indicates the total number of construction lines; T L Indicates the service life of the construction line; C l Indicates the construction cost per unit length of line; L l Indicates the length of the lth construction line.

[0051] The electricity purchase and sales of the integrated energy system can be expressed as

[0052]

[0053] Where: represents the amount of electricity purchased and sold by the integrated energy system α to the distribution network during the tth operating period; N IES is the total number of integrated energy systems participating in the game.

[0054] The electricity purchase and sales prices set by the distribution network should be limited to a certain range, which can be expressed as

[0055]

[0056] Where: They are the upper and lower limits of the power purchase price of the distribution network; They are the upper and lower limits of the electricity sales price of the distribution network.

[0057] To avoid the situation where the leading distribution network always raises the electricity price to maximize its own benefits Set it to the highest level and increase the electricity purchase price To set the lowest price, it is necessary to constrain the average price of electricity purchase and sale, which can be expressed as

[0058]

[0059]

[0060] Where: are the average value constraints of the electricity sales and purchase prices of the distribution network respectively.

[0061] This model takes minimizing network loss as an example and establishes a power flow model based on Distflow, which can be expressed as

[0062]

[0063] Where: P ij and Q ij is the active and reactive power of the branch; p i and q i is the active and reactive power of the node; v ij is the square of the branch voltage amplitude; l ij is the square of the branch current amplitude; N is the set of nodes; i and j represent nodes; ε is the set of branches, (i, j) represents a branch; r ij and l ij is the resistance and reactance of the branch; · and Indicates the lower and upper bounds of the variable ·.

[0064] (2) Distributed power generation operator operation model

[0065] The energy conversion process for wind power generation is from wind energy to mechanical energy and then to electrical energy. The wind turbine performs the conversion from wind energy to mechanical energy, while the generator performs the conversion from mechanical energy to electrical energy. During wind power generation, the wind first drives the blades of the wind turbine, then uses a speed increaser to increase the speed, and finally the generator generates the required electrical energy.

[0066] The output of a wind turbine is closely related to wind speed. When the natural wind speed is less than the cut-in wind speed, it is difficult to drive the rotor blades to rotate, and the wind turbine output power is zero. When the wind speed is greater than the cut-in wind speed but less than the rated wind speed, the wind turbine output power increases with the increase in wind speed. When the wind speed is greater than the rated wind speed but less than the cut-out wind speed, the wind turbine output power remains at the rated value. When the wind speed is greater than the cut-out wind speed, the wind turbine is shut down to prevent damage to the wind turbine, and the wind turbine output power is zero.

[0067] The relationship between the output power of a wind turbine and the natural wind speed is:

[0068]

[0069] Where: Indicates the output power of wind turbine generator set; Indicates the rated output power of the wind turbine; v indicates the natural wind speed; v in Indicates the wind speed at which the wind turbine is cut in; v N Indicates the rated wind speed of the wind turbine; v out Indicates the wind speed at which the wind turbine generator set cuts out.

[0070] Photovoltaic power generation converts solar energy into electrical energy, a process accomplished by solar cell modules. This process involves sunlight striking a solar cell module, which consists of solar cells connected in series. The solar cells then generate the required electrical energy through the photoelectric effect.

[0071] The output power of a photovoltaic power generation device is related to the intensity of natural sunlight. Within a certain range, the output power of a photovoltaic power generation device is directly proportional to the light intensity. When the light intensity is less than the rated light intensity, the output power of the photovoltaic power generation device increases proportionally with the increase in light intensity. When the light intensity is greater than the rated light intensity, the output power of the photovoltaic power generation device remains at the rated value.

[0072] The relationship between the output power of photovoltaic power generation device and the natural light intensity is:

[0073]

[0074] Where: Indicates the output power of photovoltaic power generation device; Indicates the rated output power of the photovoltaic power generation device; s indicates the light intensity; s N Indicates the rated light intensity.

[0075] Comprehensively consider the coordinated planning of wind power and photovoltaic power to achieve wind and solar complementarity. Distributed power generation operators hope to reduce construction and operation and maintenance costs and transaction expenses with energy storage operators, increase electricity sales revenue, and configure the capacity of distributed power generation with the goal of maximizing their own profits. The objective function is

[0076] maxπ DG =π dg,sel -π dg,con -π dg,ope (twenty two)

[0077] Where: π DG represents the total revenue of the distributed power generation operator; π dg,sel Represents the revenue from transactions between distributed generation operators and distribution networks; π dg,con Represents the construction investment cost of distributed power generation; π dg,ope Represents the operation and maintenance cost of distributed power generation.

[0078]

[0079] Where: T represents the total number of operating hours; represents the actual output of the wind turbine in the tth operating period; Indicates the actual output of photovoltaic power in the tth operating period.

[0080]

[0081] Where: Indicates the rated power of a single fan; Indicates the construction cost of the wind turbine per unit capacity; N WT Indicates the number of fans configured; Indicates the rated power of a single photovoltaic cell; Represents the photovoltaic construction cost per unit capacity; N PV Indicates the number of photovoltaic configurations; Indicates the service life of the fan; represents the service life of photovoltaic panels; r represents the discount rate.

[0082]

[0083] Where: Indicates the operation and maintenance cost of wind turbine per unit electricity; Represents the operation and maintenance cost per unit of photovoltaic power generation.

[0084] The configuration quantity constraint of distributed power generation can be expressed as:

[0085]

[0086] Where: Indicates the lower limit of the number of fans configured; Indicates the upper limit of the number of fans to be configured; Indicates the lower limit of photovoltaic configuration quantity; Indicates the upper limit of the number of PV configurations.

[0087] The output constraint of distributed generation can be expressed as:

[0088]

[0089] Where: represents the planned output of the wind turbine in the tth operating period; It represents the planned PV output during the t-th operating period.

[0090] (3) Energy Storage Operator Operation Model

[0091] Energy storage operators hope to increase transaction revenue with distributed power generation operators and peak-valley arbitrage income from low storage and high generation, reduce the construction and operation and maintenance costs of energy storage devices, and configure the capacity of energy storage devices with the goal of maximizing their own profits. The objective function is:

[0092] maxπ s =π s,bus +π s,ptv +π s,sub -π s,con -π s,ope (30)

[0093] Where: π s represents the revenue of energy storage operators; π s,bus Represents the transaction income with distributed power producers; π s,ptv represents the arbitrage income from low deposits and high issuance; s,sub represents government subsidies; π s,con represents the construction cost of the energy storage device; π s,ope Represents the operation and maintenance cost of the energy storage device.

[0094] π s,bus =π dg,bus (31)

[0095]

[0096] Where: represents the discharge power of the energy storage operator for peak-valley arbitrage during the t-th operating period; It represents the charging power of the energy storage operator's peak-valley arbitrage during the t-th operating period.

[0097]

[0098] Where: ρ s,sub Indicates government subsidy per unit power.

[0099]

[0100] Where: ρ s,con represents the unit capacity construction cost of the energy storage device; P s Indicates the capacity of a single energy storage device; N s Indicates the number of installed energy storage devices; T s Indicates the service life of the energy storage device.

[0101] π s,ope =βπ s,con (35)

[0102] Where: β represents the ratio coefficient between the operation and maintenance cost and the construction cost of the energy storage device.

[0103]

[0104] Where: represents the storage capacity of the energy storage device at the end of the t-1th operating period; represents the storage capacity of the energy storage device at the end of the t-th operating period; η dch Represents the discharge efficiency of the energy storage device; η ch Indicates the charging efficiency of the energy storage device.

[0105] The state of charge of the energy storage device in the tth operating period is:

[0106]

[0107] Where: Indicates the state of charge of the energy storage device at the end of the t-th operating period.

[0108] The number of energy storage devices is constrained as follows:

[0109] N s,min ≤N s ≤N s,max (39)

[0110] Where: N s,max Indicates the upper limit of the number of energy storage installations; N s,min Indicates the lower limit of the number of energy storage installations.

[0111] SOC s,min ≤SOC s,t ≤SOC s,max (40)

[0112] Where: SOC s,maxIndicates the upper limit of the state of charge of the energy storage device; SOC s,min Indicates the lower limit of the energy storage device's state of charge.

[0113] The charge and discharge power constraints of the energy storage device are:

[0114]

[0115] Where: Indicates the maximum charging power of the energy storage device; Indicates the maximum discharge power of the energy storage device.

[0116] The charge and discharge state constraints of the energy storage device are:

[0117]

[0118] Where: Represents the charging state variable of the energy storage device; Represents the discharge state variable of the energy storage device.

[0119] (4) Demand response operation model

[0120] Reducible load refers to the load that users on the demand side can adjust based on real-time energy price signals and incentive policies, such as indoor lighting brightness, air conditioning mode, and heating temperature. Integrated energy system operators can negotiate with demand-side users and provide certain load reduction compensation to encourage users to actively reduce load demand within an acceptable range during peak energy consumption periods, thereby achieving load peak reduction. The electric load model can be expressed as:

[0121]

[0122] Where: is the transferred electric load of the integrated energy system at time t; is the reduced electric load of the integrated energy system at time t; Cost of transferring electrical loads; Cut costs for electrical loads.

[0123] Demand response electric load of the integrated energy system at time t Mainly based on the forecast of electric load Transferable electrical load and reduce electrical load Composition can be expressed as:

[0124]

[0125] The constraints that can be shifted and curtailed can be expressed as:

[0126]

[0127] Where: is the maximum transferable electrical load; is the maximum transferable electrical load; The maximum electrical load that can be reduced.

[0128] The electricity transactions between the integrated energy system cooperation alliance must ensure that the transaction volume is within the limit and the transaction volume is equal, which can be expressed as:

[0129] Q min ≤Q i,j,t ≤Q max (49)

[0130] Where: Q max is the maximum interaction limit between integrated energy systems; Q min It is the minimum interaction limit between integrated energy systems.

[0131] The electric power balance constraint can be expressed as:

[0132]

[0133] 3. Hybrid game of power generation, grid, load and storage

[0134] The hierarchical scheduling framework constructed by the present invention is divided into two stages: the first stage is the Stackelberg game (master-slave game), and the second stage is the cooperative game.

[0135] Hybrid game negotiation framework Figure 2 As shown, the system first initializes the population of energy trading volumes and the number of iterations of velocity. It then updates the displacement and velocity of each particle and calculates its fitness. If the difference between the previous and current energy trading volumes is within a certain threshold, convergence (P1) is reached. Otherwise, the historical optimal displacement of each particle and the historical optimal displacement of all particles are updated, and iterations continue. After P1, the objective functions of different integrated energy systems are iterated using ADMM decomposition, and deep reinforcement learning is used to solve them and obtain the P2P energy trading price.

[0136] (1) Master-slave game solution algorithm

[0137] In Stackelberg game optimization models, KKT conditions or heuristic algorithms are often used. In optimization theory, KKT conditions are necessary for optimal solutions to nonlinear programming problems. They generalize the Lagrange multiplier method to inequality constraints. Because the underlying model, the IES cooperative alliance, contains 0-1 variables, KKT conditions cannot be converted into constraints. Instead, heuristic algorithms can be used to solve this problem.

[0138] (2) Cooperative game solving algorithm

[0139] After realizing ADMM decomposition, the present invention adopts deep reinforcement learning algorithm to solve the P2P energy transaction price, so as to achieve fair and reasonable distribution of cooperative profits.

[0140] 4. Case analysis

[0141] (1) Hybrid game optimization results

[0142] by Figure 3 The framework presented here considers two game models. Option A considers a master-slave game between the distribution network and the IES Alliance, a cooperative game between IES Alliance members, and demand-side response, i.e., the method proposed in this paper. Option B considers a master-slave game between the distribution network and the IES Alliance, ignores the cooperative game between IES Alliance members, and considers demand-side response.

[0143] Tables 1 through 3 provide key parameters for integrated energy parks. For Integrated Energy Park 1, the Juhua Group Source-Grid-Load-Storage Silicon-Fluorine New Materials Industrial Cluster Project is used as an example. The project is located in the Chemical Industrial Park in the old urban area of ​​Yumen City, Jiuquan City. Regarding load, upon completion, the projected maximum load is 2.207 million kilowatts, with an annual utilization of 8,000 hours and an annual electricity consumption of 17.66 billion kWh. Regarding power generation, the total installed capacity of supporting renewable energy is 4.95 million kilowatts, including 3.465 million kilowatts of wind power, 3,068 hours of annual utilization, and an annual power generation of 10.63 billion kWh; and 1.485 million kilowatts of photovoltaic power, 1,960 hours of annual utilization, and an annual power generation of 2.91 billion kWh. The total annual power generation from renewable energy is 13.54 billion kWh. Regarding power generation and consumption matching, with the deployment of 750,000 kilowatts / 3 million kWh of energy storage, the renewable energy utilization rate reaches over 95%, and the self-generated and self-consumed electricity from renewable energy reaches 12.89 billion kWh.

[0144] Table 1 Key parameters of integrated energy park 1

[0145]

[0146] Table 2 Key parameters of integrated energy park 2

[0147]

[0148] Table 3 Key parameters of integrated energy park 3

[0149]

[0150] In the two-stage solution process constructed by the model, the master-slave game stage determines the electricity interaction price between the integrated energy system (IES) cooperative alliance and the distribution network and IES cooperative alliance members.

[0151] Table 4 Profit comparison before and after the alliance

[0152]

[0153] Table 4 shows the profit potential of different scenarios within the framework of interaction between multiple integrated energy systems and distribution networks. By incorporating renewable energy power plants, energy storage operators, and demand-side operators into the same collaborative alliance, the costs of each integrated energy system are significantly reduced, by approximately 50%. This is because energy storage provides the spatial and temporal transfer of electricity, enabling optimal energy distribution.

[0154] Figure 4 This figure shows electricity prices for different scenarios within the framework of interaction between multiple integrated energy systems and the distribution network. The electricity purchase price set by Scenario 1 is lower than that set by Scenario 2. This is because Scenario 1 considers cooperative game play between IES alliance members and within IES itself. This provides new ways for source, load, and storage entities to mitigate the risk of load fluctuations, reduce dependence on the distribution network, and create more room for negotiation.

[0155] Figure 5 The figure shows the interactive electricity prices between integrated energy parks. As can be seen, the parks exchange electricity through cooperative bargaining, and the interactive electricity prices are all lower than the off-grid electricity prices. To ensure the interests of cooperation, the IES cooperative alliance must set electricity prices within the purchase and sale prices established in Phase 1. This ensures that IES members will choose to cooperate through inter-IES electricity trading. Furthermore, because each IEM cooperative alliance member has the ability to independently set prices, this not only protects the interests of all members but also provides them with better bargaining power with distribution network operators.

[0156] Figure 6 This diagram illustrates the operation of Integrated Energy Park 1 within a hybrid game framework. During periods 0-4, because wind power output exceeds load demand, Integrated Energy System 1 stores the surplus energy as energy storage. During periods 4-8, due to low wind power output and virtually no photovoltaic generation, Integrated Energy Park 1 can utilize energy storage to meet load demand. During periods 11-15, Park 1's renewable energy output exceeds load demand. While meeting energy storage requirements, the excess electricity is sold to Integrated Energy Parks 2 and 3, achieving optimal energy distribution and meeting the overall alliance's electricity needs.

[0157] Table 5 Profit distribution of cooperative alliance

[0158] Alliance Members Profit / yuan Integrated Energy System No. 1 After the cooperation 185737 Integrated Energy System No. 2 After the cooperation 164403 Integrated Energy System No. 3 After the cooperation 176463

[0159] Table 5 shows the profit distribution of multiple integrated energy systems after cooperation. The profits account for one-third of the cooperative alliance, achieving a balanced distribution of interests and verifying the effectiveness of the Nash negotiation proposed in this article.

[0160] In summary, the hybrid game model constructed in the present invention can formulate the purchase and sale electricity prices of the integrated energy system cooperative alliance and the electricity trading prices between system cooperative alliances. While ensuring the coordinated operation of the distribution network and the IES cooperative alliance, it also realizes the cooperation and profit distribution of the IES cooperative alliance. The constructed model effectively improves the benefits of each IES.

Claims

1. A multi-agent optimization scheduling method for source, grid, load and storage based on hybrid game theory, characterized by: The following steps are involved: Step 1: Construct a hybrid game model of the source-grid-load-storage framework considering power trading: The distribution network operators and the integrated energy system cooperative alliance adopt a master-slave game. The distribution network operators are the leaders, who influence the dispatching strategy of the integrated energy system by setting energy transaction prices to maximize profits. The integrated energy system is the follower, who accepts the prices set by the distribution network and changes its energy consumption pattern to reduce its own operating costs. At the same time, different integrated energy systems use cooperative game theory to exchange energy. When there are differences in renewable energy generation and load demand at different times among integrated energy systems, the integrated energy systems achieve energy sharing and fair distribution of benefits through cooperation. Each integrated energy system uses Nash negotiation to determine the electricity transaction volume and transaction price. Step 2: Construct a multi-agent optimization scheduling model for source, grid, load and storage: Taking into account the interests of distributed power generation operators, distribution network operators, energy storage operators, and the demand side, a multi-agent optimization scheduling model for source, grid, load, and storage is constructed. This model includes a distribution network operation model, a distributed power generation operator operation model, an energy storage operator operation model, and a demand response operation model. Step 3: Solve the hybrid game of source, grid, load and storage The particle swarm optimization algorithm is combined with a deep learning algorithm based on alternating direction multiplier decomposition. The particle swarm optimization algorithm solves the master-slave game, and the deep learning algorithm based on alternating direction multiplier decomposition solves the cooperative game. The two jointly solve the constructed hybrid game model and obtain the multi-agent optimization scheduling parameters of the source, grid, load and storage. Specifically: First, the population of electric energy trading volume and the number of iterations of speed are initialized. Then the displacement and speed of each particle are updated and the fitness is calculated. If the difference between the previous electric energy trading volume and the current electric energy trading volume is within the set threshold and convergence is reached, the master-slave game is solved and ended. Otherwise, the historical optimal displacement of each particle and the historical optimal displacement of all particles are updated and iterations continue. After the master-slave game is solved, the objective functions of different integrated energy systems are iterated again using alternating direction multiplier decomposition, and the deep reinforcement learning method is used to solve and obtain the P2P energy trading price.

2. The source-grid-load-storage multi-agent optimization scheduling method based on hybrid game theory according to claim 1 is characterized in that: The distribution network operation model in step 2 is specifically: Distribution network operators comprehensively consider the revenue from electricity purchase and sales, grid construction costs, and failure costs to ensure the safe and economical operation of the distribution network; the objective function is: maxp G =π G2IES +p G2PN -p O&M (1) Where: π G represents the profit of the distribution network; π G2IES represents the transaction income between the distribution network and the integrated energy system; π G2PN Represents the transaction income between the distribution network and the main grid; π O&M represents the line construction cost of the distribution network; Where: ρ t G,sel represents the price of electricity sold by the distribution network to the integrated energy system during the t-th operating period; ρ t G,buy represents the price of electricity purchased by the distribution network from the integrated energy system during the t-th operating period; represents the amount of electricity sold by the distribution network to the integrated energy system during the t-th operating period; It represents the amount of electricity purchased from the integrated energy system by the distribution network during the t-th operating period; T is the total number of operating hours; Where: ρ t G2PN,sell represents the price of electricity sold by the distribution network to the main grid during the t-th operating period; ρ t G2PN,buy represents the price of electricity purchased by the distribution network from the main grid during the t-th operating period; represents the amount of electricity sold by the distribution network to the main grid during the t-th operating period; represents the amount of electricity purchased by the distribution network from the main grid during the t-th operating period; Where: N l Indicates the total number of construction lines; T L Indicates the service life of the construction line; C l Indicates the construction cost per unit length of line; L l represents the length of the first construction line; r represents the discount rate, which refers to converting a certain amount of money in the future into an equivalent price today; Electricity purchase and sales by the Integrated Energy System Cooperation Alliance And is expressed as: Where: and represents the amount of electricity purchased and sold by the integrated energy system cooperation alliance from the distribution network during the tth operating period; and represents the amount of electricity purchased and sold by the integrated energy system α to the distribution network during the tth operating period; N IES is the total number of integrated energy systems participating in the game; The electricity purchase and sales prices set by the distribution network should be limited to a certain range, expressed as: Where: ρ t G,buy and ρ t G,sel are the purchase and sale prices of electricity from the distribution network, respectively; and They are the upper and lower limits of the power purchase price of the distribution network; and They are the upper and lower limits of the electricity sales price of the distribution network; The average value constraint of electricity purchase and sales price is expressed as: Where: and are the average value constraints of the electricity sales and purchase prices of the distribution network respectively.

3. The source-grid-load-storage multi-agent optimization scheduling method based on hybrid game theory according to claim 2 is characterized in that: The distribution network operation model is established with the goal of minimizing network loss. The power flow model based on Distflow is expressed as follows: Where: P ij and Q ij is the active and reactive power of branch (i, j); p i and q i is the active and reactive power of the node; v i and v j is the voltage of node i, j; v ij is the square of the branch voltage amplitude; l ij is the square of the branch current amplitude; N is the set of nodes; i and j represent nodes; ε is the set of branches, (i, j) represents a branch; r ij and x ij is the resistance and reactance of the branch; · and Indicates the lower and upper bounds of the variable ·.

4. The source-grid-load-storage multi-agent optimization scheduling method based on hybrid game theory according to claim 1 is characterized in that: The distributed power operator operation model in step 2 is specifically as follows: The relationship between the output power of a wind turbine and the natural wind speed is: Where: P t WT Indicates the output power of wind turbine generator set; Indicates the rated output power of the wind turbine; v indicates the natural wind speed; v in Indicates the wind speed at which the wind turbine is cut in; v N Indicates the rated wind speed of the wind turbine; v out Indicates the wind speed at which the wind turbine generator set is cut out; The relationship between the output power of photovoltaic power generation device and the natural light intensity is: Where: P t PV Indicates the output power of photovoltaic power generation device; Indicates the rated output power of the photovoltaic power generation device; s represents the light intensity; s N Indicates rated light intensity; Distributed power generation operators configure the capacity of distributed power generation with the goal of maximizing their own profits. The objective function is: maxp DG =π dg,sel -p dg,con -p dg,ope (22) Where: π DG represents the total revenue of the distributed power generation operator; π dg,sel Represents the revenue from transactions between distributed generation operators and distribution networks; π dg,con Represents the construction investment cost of distributed power generation; π dg,ope represents the operation and maintenance cost of distributed power generation; Where: T represents the total number of operating hours; represents the actual output of the wind turbine in the tth operating period; represents the actual output of photovoltaic power in the tth operating period; ρ t G,buy is the electricity purchase price of the distribution network; Where: Indicates the construction cost of the wind turbine per unit capacity; N WT Indicates the number of fans configured; Indicates the rated power of a single photovoltaic cell; Represents the photovoltaic construction cost per unit capacity; N PV Indicates the number of photovoltaic configurations; Indicates the service life of the fan; represents the service life of photovoltaic panels; r represents the discount rate; Where: Indicates the operation and maintenance cost of wind turbine per unit electricity; Indicates the operation and maintenance costs per unit of photovoltaic power generation; The configuration quantity constraint of distributed power generation is expressed as: Where: Indicates the lower limit of the number of fans configured; Indicates the upper limit of the number of fans to be configured; Indicates the lower limit of photovoltaic configuration quantity; Indicates the upper limit of photovoltaic configuration quantity; The output constraint of distributed generation is expressed as: Where: represents the planned output of the wind turbine in the tth operating period; It represents the planned PV output during the t-th operating period.

5. The source-grid-load-storage multi-agent optimization scheduling method based on hybrid game theory according to claim 1 is characterized in that: The energy storage operator operation model in step 2 is specifically as follows: Energy storage operators configure the capacity of energy storage devices with the goal of maximizing their own profits. The objective function is: maxp s =π s,bus +p s,ptv +p s,sub -p s,con -p s,ope (30) Where: π s represents the revenue of energy storage operators; π s,bus Represents the transaction income with distributed power producers; π s,ptv represents the arbitrage income from low deposits and high issuance; s,sub represents government subsidies; π s,con represents the construction cost of the energy storage device; π s,ope represents the operation and maintenance cost of the energy storage device; π s,bus ＝π dg,bus (31) Where: P t dch2 P represents the discharge power of the energy storage operator for peak-valley arbitrage during the t-th operating period; t ch2 represents the charging power of the energy storage operator's peak-valley arbitrage during the t-th operating period; ρ t g,set is the energy storage charging and discharging cost, π dg,bus Energy storage operation and maintenance costs; Where: ρ s,sub represents the government subsidy per unit power; P t ch1 Charging power for energy storage; Where: ρ s,con represents the unit capacity construction cost of the energy storage device; P s Indicates the capacity of a single energy storage device; N s Indicates the number of installed energy storage devices; T s Indicates the service life of the energy storage device; π s,ope ＝βπ s,con (35) Where: β represents the ratio coefficient between the operation and maintenance cost of the energy storage device and the construction cost; Where: represents the storage capacity of the energy storage device at the end of the t-1th operating period; P t s,sur represents the storage capacity of the energy storage device at the end of the t-th operating period; η dch Indicates the discharge efficiency of the energy storage device; η ch Indicates the charging efficiency of the energy storage device; P t dch and P t ch are the discharge power and charging power of energy storage at time t respectively; The state of charge of the energy storage device in the tth operating period is: Where: Indicates the state of charge of the energy storage device at the end of the t-th operating period; The number of energy storage devices is constrained as follows: N s,min ≤N s ≤N s,max (39) Where: N s,max Indicates the upper limit of the number of energy storage installations; N s,min Indicates the lower limit of the number of energy storage installations; SOC s,min ≤SOC s,t ≤SOC s,max (40) Where: SOC s,max Indicates the upper limit of the state of charge of the energy storage device; SOC s,min Indicates the lower limit of the state of charge of the energy storage device; The charge and discharge power constraints of the energy storage device are: Where: Indicates the maximum charging power of the energy storage device; Indicates the maximum discharge power of the energy storage device; Represents the charging state variable of the energy storage device; The energy storage device charge and discharge state constraints are:

6. The source-grid-load-storage multi-agent optimization scheduling method based on hybrid game theory according to claim 1 is characterized in that: The demand response operation model in step 2 is specifically as follows: The electric load model is expressed as: Where: π DR for the benefits of demand-side response; is the transferred electric load of the integrated energy system at time t; is the reduced electric load of the integrated energy system at time t; ρ t tran Cost of transferring electrical loads; ρ t cut Cost reduction for electrical loads; Demand response electric load of the integrated energy system at time t By predicting the electric load Transferable electrical load and reduce electrical load Composition, expressed as: The constraints that can be shifted and reduced are expressed as: Where: is the maximum transferable electrical load; is the maximum transferable electrical load; The maximum electrical load that can be reduced; The electricity transactions between the integrated energy system cooperation alliance must ensure that the transaction volume is within the limit and the transaction volume is equal, which can be expressed as: Q min ≤Q i,j,t ≤Q max (49) Where: Q max is the maximum interaction limit between integrated energy systems; Q min is the minimum interaction limit between integrated energy systems; Q i,j,t is the electric energy interaction between nodes i and j in the integrated energy system at time t; The electric power balance constraint is expressed as: Where: and They are electricity purchased by the integrated energy system, energy storage discharge, wind power output, photovoltaic output, demand-side response and energy storage charging.

Images