A multi-region integrated energy system three-layer game optimization scheduling method, system, device and medium considering mobile energy storage willingness transmission

Abstract

This application provides a three-layer game-theoretic optimization scheduling method, system, equipment, and medium for multi-regional integrated energy systems that considers the transmission of mobile energy storage willingness, relating to the field of energy dispatching decision-making technology. The method includes: acquiring system operation data and vehicle user characteristic data; constructing a vehicle user response willingness transmission model, generating the user's final participation willingness through saturated incentive prediction, load correlation analysis, and graph convolutional networks; establishing a three-layer game-theoretic scheduling model including master-slave game, cooperative game, and group game, solving for the incentive price and dispatching decisions for each region; and employing a game-theoretic reinforcement learning algorithm, treating the dispatching center as an intelligent agent, to output the optimal dispatching strategy. This application can achieve accurate characterization of vehicle user response behavior, enhanced inter-regional collaborative operation capabilities, and reduced overall system operating costs.

Description

Technical Field

[0001] This application relates to the field of energy dispatch decision technology, and in particular to a three-layer game optimization dispatch method, system, equipment and medium for multi-region integrated energy systems that considers the transmission of mobile energy storage intentions. Background Technology

[0002] With the continuous advancement of the construction of new power systems and new energy systems, electric vehicles and hydrogen fuel cell vehicles, as typical mobile energy storage resources, are gradually transforming from traditional energy loads into important demand response resources with flexible adjustment capabilities. Multi-regional integrated energy systems, through the complementarity of multiple energy sources such as electricity, heat, and hydrogen, and inter-regional energy sharing, can effectively improve energy utilization efficiency and system operation economy. However, in scenarios involving large-scale integration of mobile energy storage, existing dispatching methods still have the following problems: (1) Most existing vehicle user response modeling treats users as independent individuals and focuses on analyzing the impact of factors such as incentive price, charging demand, and battery status on individual intentions. However, it ignores the intention propagation effect generated by social connections, behavioral imitation, and information interaction among users, resulting in insufficient accuracy of user response modeling.

[0003] (2) Although existing multi-regional integrated energy system optimization scheduling models take into account the coordination of the interests of multiple stakeholders, most of them only focus on the local game relationship between the dispatch center and each region, or between the region and the user, and fail to uniformly depict the multi-level coupling relationship between the dispatch center, the regional integrated energy system and the vehicle user group.

[0004] (3) In a multi-regional integrated energy system, the vehicle user response process, the regional collaborative operation process and the upper-level incentive decision-making process are coupled with each other. The system has significant nonlinear, time-varying and non-analytical characteristics, and traditional optimization algorithms are difficult to achieve efficient solutions.

[0005] Therefore, there is an urgent need to propose a multi-regional integrated energy system optimization and scheduling method that can simultaneously consider the mobile energy storage user willingness transmission mechanism, multi-regional collaborative operation mechanism, and dynamic incentive optimization mechanism, so as to improve the accuracy of user response modeling, enhance multi-regional collaborative scheduling capability, and reduce the overall system operating cost. Summary of the Invention

[0006] This application provides a three-layer game-theoretic optimization scheduling method, system, equipment, and medium for multi-region integrated energy systems that considers the transmission of mobile energy storage intentions. It is used to solve problems in the prior art such as overly static modeling of vehicle user response intentions, insufficient characterization of user group propagation behavior, incomplete modeling of multi-subject collaborative decision-making in multi-region integrated energy systems, and difficulty in scheduling solutions. This enables accurate characterization of vehicle user response behavior, enhanced inter-regional collaborative operation capabilities, and reduced overall system operating costs.

[0007] Firstly, this application provides a three-layer game-theoretic optimization scheduling method for a multi-regional integrated energy system that considers the transmission of mobile energy storage intentions, the method comprising: The system acquires operational status data and vehicle user characteristic data for a multi-regional integrated energy system. The operational status data includes electricity, heat, and hydrogen load demand, new energy output, energy storage status, and the energy purchase price of the upstream power grid in each region. The vehicle user characteristic data includes vehicle type, dispatchable capacity, real-time energy price, energy station congestion level, and time-series data of user load characteristics. Based on the vehicle user characteristic data, a vehicle user response intention transmission model is constructed. The vehicle user response intention transmission model determines the initial response intention of individual users through saturation incentive prediction. A load characteristic correlation analysis based on the Hilbert-Schmidt independence criterion is used to construct a user intention influence topology. A graph convolutional network is used to iteratively transmit intentions on the user intention influence topology to obtain the final participation intention level of the user group. Based on the final participation willingness level and the operational status data, a three-layer game scheduling model for a multi-regional integrated energy system is established. The three-layer game scheduling model includes a master-slave game model between the scheduling center and the multi-regional integrated energy system, a cooperative game model between the integrated energy systems in each region, and a group game model based on willingness transmission among vehicle users in the region. The scheduling strategies of each subject are obtained by solving the three-layer game scheduling model. The scheduling strategies include incentive price, power purchase capacity, inter-regional interactive power, and equipment operation plan within each region. The game-reinforcement learning algorithm is used to solve the three-layer game scheduling model. The scheduling center is regarded as an intelligent agent, and the response behavior of the multi-region integrated energy system and its internal vehicle users is regarded as the environment. With the goal of minimizing the system operating cost, the optimal incentive price and regional scheduling decision are output.

[0008] Secondly, this application provides a three-layer game-theoretic optimization scheduling system for a multi-regional integrated energy system that considers the transmission of mobile energy storage intentions, the system comprising: The data acquisition module is configured to acquire the operation status data and vehicle user characteristic data of the multi-regional integrated energy system. The operation status data includes the electricity, heat, and hydrogen load demand, new energy output, energy storage status, and the energy purchase price of the upper-level power grid in each region. The vehicle user characteristic data includes vehicle type, dispatchable capacity, real-time energy price, energy station congestion level, and time-series data of user load characteristics. The intention transmission module is configured to construct a vehicle user response intention transmission model based on the vehicle user characteristic data. The vehicle user response intention transmission model determines the initial response intention of individual users through saturation incentive prediction, constructs a user intention influence topology map by using load characteristic correlation analysis based on the Hilbert-Schmidt independence criterion, and uses a graph convolutional network to iteratively transmit intentions on the user intention influence topology map to obtain the final participation intention level of the user group. The three-layer game module is configured to establish a three-layer game scheduling model for a multi-regional integrated energy system based on the final participation willingness level and the operating status data. The three-layer game scheduling model includes a master-slave game model between the scheduling center and the multi-regional integrated energy system, a cooperative game model between the integrated energy systems in each region, and a group game model based on willingness transmission among vehicle users in the region. The scheduling strategy of each subject is obtained by solving the three-layer game scheduling model. The scheduling strategy includes incentive price, power purchase capacity, inter-regional interactive power, and equipment operation plan within each region. The model solving module is configured to use a game-reinforcement learning algorithm to solve the three-layer game scheduling model. The scheduling center is regarded as an intelligent agent, and the response behavior of the multi-region integrated energy system and its internal vehicle users is regarded as the environment. With the goal of minimizing the system operating cost, the module outputs the optimal incentive price and regional scheduling decision.

[0009] Thirdly, embodiments of this application provide an electronic device, including: at least one processor and a memory; the memory stores computer execution instructions; the at least one processor executes the computer execution instructions stored in the memory, causing the at least one processor to execute the three-layer game optimization scheduling method for multi-region integrated energy systems considering the transmission of mobile energy storage intentions as described in the first aspect and various possible designs of the first aspect.

[0010] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer-executable instructions. When a processor executes the computer-executable instructions, it implements the three-layer game optimization scheduling method for multi-region integrated energy systems that considers the transmission of mobile energy storage intentions, as described in the first aspect and various possible designs of the first aspect.

[0011] Fifthly, embodiments of this application provide a computer program product, including a computer program that, when executed by a processor, implements the three-layer game optimization scheduling method for multi-region integrated energy systems that considers the transmission of mobile energy storage intentions, as described in the first aspect and various possible designs of the first aspect.

[0012] The three-layer game-theoretic optimization scheduling method, system, equipment, and medium for multi-regional integrated energy systems that consider the transmission of mobile energy storage intentions provided in this application have at least the following beneficial effects: Compared with existing methods, this application achieves accurate characterization of vehicle user response behavior, enhanced inter-regional collaborative operation capabilities, and reduced overall system operating costs through steps such as constructing a multi-regional integrated energy system architecture, modeling vehicle user response intentions, conducting user association analysis based on HSIC correlation, constructing a user intention transmission mechanism based on graph convolutional neural networks, establishing a three-layer game scheduling model for the multi-regional integrated energy system, and solving the problem based on game theory and reinforcement learning. Specifically, this application not only considers the individual response characteristics of mobile energy storage users such as electric vehicles and hydrogen fuel cell vehicles under incentives, but also further characterizes the intention propagation effect caused by similar load characteristics, information interaction, and behavioral imitation among users. Simultaneously, it constructs a three-layer game-based collaborative optimization framework among the dispatch center, each regional integrated energy system, and vehicle user groups within the region, and combines reinforcement learning to achieve adaptive optimization of dynamic incentive prices. This application can more realistically reflect the response behavior of mobile energy storage user groups and the collaborative operation characteristics of multi-regional integrated energy systems, thereby effectively improving the demand response resource aggregation capability, enhancing the regional collaborative scheduling level, and reducing the overall system operating costs, demonstrating strong applicability and practical value. Attached Figure Description

[0013] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0014] Figure 1 A schematic diagram of a three-layer game optimization scheduling method for a multi-region integrated energy system that considers the transmission of mobile energy storage intentions, provided for an embodiment of this application; Figure 2 A flowchart of a three-layer game optimization scheduling method for a multi-region integrated energy system that considers the transmission of mobile energy storage intentions is provided for embodiments of this application; Figure 3 A multi-region integrated energy system structure diagram provided for embodiments of this application; Figure 4 This is a schematic diagram illustrating the correlation analysis of vehicle user intentions provided in an embodiment of this application; Figure 5 This is a schematic diagram illustrating the process of transmitting vehicle user intentions in an embodiment of this application. Figure 6 The new energy output and load diagrams provided in the embodiments of this application; Figure 7 The running result diagram provided for the embodiments of this application; Figure 8The inter-regional average electrical exchange power matrix diagram provided for embodiments of this application; Figure 9 The user participation rate distribution chart considering intention transmission is provided for the embodiments of this application; Figure 10 This application provides a user participation rate distribution chart for users who did not communicate their intentions in an embodiment of the application. Figure 11 The graph showing the incremental change in participation rate before and after intention transmission is provided in the embodiments of this application; Figure 12 The structural diagram of a three-layer game-theoretic optimization scheduling system for a multi-regional integrated energy system that considers the transmission of mobile energy storage intentions is provided in the embodiments of this application.

[0015] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concepts of this application to those skilled in the art through reference to specific embodiments. Detailed Implementation

[0016] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of systems and methods consistent with some aspects of this application as detailed in the appended claims.

[0017] The collection, storage, use, processing, transmission, provision, and disclosure of financial data or user data involved in the technical solution of this application all comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0018] It should be noted that in the embodiments of this application, certain software, components, models and other existing solutions in the industry may be mentioned. These should be regarded as exemplary and are only intended to illustrate the feasibility of implementing the technical solution of this application. However, it does not mean that the applicant has used or necessarily used the solution.

[0019] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0020] This application provides a three-layer game-theoretic optimization scheduling method for multi-regional integrated energy systems that considers the transmission of mobile energy storage willingness. The method principle diagram provided in this application embodiment is shown below. This method adopts a three-layer progressive game architecture of dispatch center – regional integrated energy system – vehicle user group. The three layers achieve bidirectional interaction through incentive prices, regional operating status, dispatchable power, user willingness, and aggregated power.

[0021] The first layer is a master-slave game between the dispatch center and the integrated energy systems in each region. In this layer, the dispatch center is the upper-level leader and the integrated energy systems in each region are the lower-level followers. Dynamic incentives and global scheduling are achieved through DDPG reinforcement learning agents, including a state perception and decision-making module and a dynamic incentive formulation (DDPG RL Agent) module.

[0022] The state awareness and decision-making module's scheduling center collects global data to construct state vectors. S t As shown in formulas (31) and (32), it includes the cluster average willingness, saturation incentive, DR schedulable power, average SOC of electric vehicles, and system load status. Based on the state-aware results, an action vector is output. a t This includes the set of differentiated incentive prices issued to each region and the power dispatch instructions for each region.

[0023] The dynamic incentive formulation module adopts the DDPG framework. It outputs deterministic actions through an Actor network, with the mapping relationship shown in the formula. The Critic network evaluates the value of each action, with the value function shown in the formula. The agent uses maximizing cumulative reward as its objective function and iteratively optimizes the network parameters to achieve adaptive dynamic incentives.

[0024] The first layer sends incentive prices and scheduling instructions to the lower layer, while simultaneously receiving regional operation status data returned by the lower layer.

[0025] The second layer involves cooperative game theory within a multi-regional integrated energy system. This layer comprises three typical regions: commercial IES, residential IES, and agricultural IES. Each region engages in cooperative game theory through energy and information interaction. The objective function is to minimize the joint operation cost of multiple regions. Decision variables include unit start-up and shutdown strategies, inter-regional power exchange, and energy output from equipment such as photovoltaic (PV), gas turbine (MT), gas boiler (GB), electrolyzer (EL), fuel cell (FC), and energy storage (ES). Inter-regional interaction is achieved based on a constructed two-way energy and information exchange network, enabling cross-regional multi-energy complementarity and surplus / deficit balance.

[0026] The second layer receives scheduling instructions from the upper layer, publishes the schedulable power demand to the lower layer, receives user preferences and aggregated schedulable power returned from the lower layer, and feeds back the overall regional operating status to the upper layer.

[0027] The third layer is the game and intention transmission mechanism among vehicle users within the region. This layer constructs an intention transmission mechanism based on a graph neural network for two heterogeneous user groups: EVs and FCEVs, and it is divided into two stages.

[0028] Phase 1: Generating Initial User Response Intent. Collecting multi-dimensional features of vehicles / regions to construct the input feature vector. X t As shown in Formula (2); input the saturated incentive prediction model based on GRU to obtain the saturated incentive, as shown in Formula (1); combine the incentive price and calculate the user's initial participation probability through the initial willingness probability model, as shown in Formula (3).

[0029] Phase 2: Willpower transmission mechanism based on graph neural networks, including: performing HSIC correlation analysis to quantify the strength of behavioral associations between users; constructing a topology graph of user willpower influence and generating an adjacency matrix representing node connection relationships. A ; By aggregating neighborhood information through the GCN network, the dynamic propagation of user intentions is realized, as shown in formula (11); by mapping interval thresholds, continuous intention values ​​are converted into five discrete intention levels (-2,-1,0,1,2), as shown in formula (12).

[0030] Ultimately, participation is determined based on the user's final intention, and the regional schedulable power and the average intention of the group are aggregated and fed back to the upper-level cooperative game layer.

[0031] Based on the above principles, combined with Figure 2 As shown, the three-layer game optimization scheduling method for multi-region integrated energy systems that considers the transmission of mobile energy storage intentions includes the following steps S10-S40.

[0032] S10: Obtain operational status data and vehicle user characteristic data of multi-regional integrated energy systems. Operational status data includes electricity, heat, and hydrogen load demand, new energy output, energy storage status, and energy purchase price from the upstream power grid in each region. Vehicle user characteristic data includes vehicle type, dispatchable capacity, real-time energy price, energy station congestion level, and time-series data of user load characteristics.

[0033] The purpose of step S10 is to construct a multi-region integrated energy system architecture. In this embodiment, a multi-region integrated energy system is constructed, consisting of a residential area integrated energy system, a commercial area integrated energy system, and an agricultural area integrated energy system. Each area includes multiple energy forms such as electricity, heat, and hydrogen, and is equipped with distributed power sources, energy conversion devices, energy storage equipment, and mobile energy storage user access facilities. Under the coordination of a unified dispatch center, cross-regional energy collaboration and optimized operation are achieved. The multi-region integrated energy system is as follows: Figure 3 As shown.

[0034] S20: Based on vehicle user characteristic data, a vehicle user response intention transmission model is constructed. The vehicle user response intention transmission model determines the initial response intention of individual users through saturation incentive prediction. A load characteristic correlation analysis based on the Hilbert-Schmidt independence criterion is used to construct a user intention influence topology graph. A graph convolutional network is used to iteratively transmit intentions on the user intention influence topology graph to obtain the final participation intention level of the user group.

[0035] In this embodiment, for the vehicle user response intention transmission model, considering the differences between EV and FCEV users in terms of energy carrier, energy consumption mode and price sensitivity, EV users and FCEV users are regarded as two heterogeneous vehicle users, and their response intention and intention transmission process are modeled under a unified incentive framework.

[0036] In some embodiments, the saturation incentive is defined as the minimum incentive price that would make all eligible vehicle users willing to participate in demand response. A GRU-based saturation incentive prediction model is proposed, which combines multi-dimensional inputs to predict the saturation incentive level in charging scenarios. User saturation incentives are influenced by factors such as vehicle type, region type, dispatchable capacity, real-time energy price, and energy station congestion. This embodiment constructs a mapping relationship from five-dimensional input feature variables to the saturation incentive level, which can be expressed as: (1) (2) in, Provide saturation incentives for vehicle users; The input feature vector; These are the trainable parameters of the model; V This indicates the vehicle type, used to distinguish between electric vehicles (EVs) and hydrogen fuel cell vehicles (FCEVs). Different vehicle types correspond to different energy carriers and price-sensitive characteristics, and their differences are characterized by parameter learning in the saturation excitation prediction model. R Indicates the region type; The schedulable capacity represents the power difference between the current schedulable demand and the available capacity. This indicates the current energy price level. The congestion level of the energy station indicates its current utilization rate. f This represents a mapping relationship.

[0037] By constructing a probability model of vehicle user response intention through a saturated incentive price, let the first... The car at any time The probability of participation is The expression is: (3) in, Let the vehicle response incentive price be set. It follows a Gaussian normal distribution.

[0038] Mapping the probability of vehicle user response intention to a discretized initial participation intention level The expression is: (4) In this system, -2 represents strong unwillingness, -1 represents somewhat unwillingness, 0 represents neutrality, 1 represents somewhat willingness, and 2 represents strong willingness. An influence factor is set based on the user's initial willingness level. For example, a very willing user will actively promote the service to other users, so the influence factor for their immediate surroundings is set to 2; for willing users, the influence factor is set to 1; for neutral users, the influence factor is set to 0; for somewhat unwilling users, the influence factor is set to -1; and for completely unwilling users, the influence factor is set to -2.

[0039] Users with similar load characteristics tend to have a stronger correlation in their willingness to participate in scheduling, and their willingness to participate is more easily transmitted among users through information exchange, behavioral imitation, and experience dissemination. Users with a high degree of willingness to participate in scheduling tend to promote their ideas to other users, thereby influencing other users' willingness to participate in scheduling. To describe this effect, a willingness transmission mechanism based on GCN is constructed in some embodiments.

[0040] Because vehicle users exhibit significant differences in charging times, refueling durations, and usage habits, their load curves typically display obvious nonlinear, multimodal, and temporally misaligned characteristics. Traditional linear correlation coefficients are insufficient to fully reflect these complex dependencies. HSIC can effectively capture complex nonlinear correlations between high-dimensional variables, making it particularly suitable for describing the complex correlation structure between vehicle user load curves. Therefore, this embodiment uses HSIC to perform correlation analysis on user load characteristics. By calculating the HSIC correlation coefficient between any two user load characteristics, the behavioral correlation strength between vehicle users is obtained, providing a basis for the subsequent construction of the intention-influence topology map. The user intention correlation analysis based on HSIC is as follows: Figure 4 As shown.

[0041] The key to HSIC computation lies in providing two positive definite kernels K and L such that existing input and output variables belonging to the input and output spaces can be mapped to their respective reproducing kernel Hilbert spaces (RKHS). The calculation formula is as follows: (5) (6) (7) (8) in, For cross covariance operator The Hilbert-Schmidt norm; For bandwidth hyperparameters, x and x' Represents random variables X Two load characteristic samples; y and y′ Represents random variables Y Two load characteristic samples; K(x,x′) and L(y,y′) Defined in X and Y The Gaussian kernel function on the x-axis is used to characterize the nonlinear similarity between samples; exp(·) represents the exponential function. and These represent the kernel functions respectively. K and L The induced regenerative nucleus Hilbert space; E This represents the expectation operator, used to statistically average the kernel function values ​​among vehicle user load characteristic samples; HSIC The correlation coefficient is the Hilbert-Schmidt independence criterion.

[0042] The load characteristics of each vehicle were selected as a sample. x Calculate each vehicle pair ( i,j HSIC correlation coefficient between ) (9) in, H ij To indicate vehicle user i With vehicle users j The correlation coefficient of load characteristics between them; x i and x j They represent the first i The and the first j Load characteristic samples of individual vehicle users.

[0043] To further describe the process of transmitting vehicle user intentions, a topology graph is constructed. Traditional models struggle to characterize the high-dimensional nonlinear dependency structure between nodes. This embodiment introduces a Generative Network (GCN) to aggregate and update information about nodes and their neighborhoods in the graph, thereby describing the dynamic transmission mechanism of vehicle user intentions in the influence network.

[0044] By abstracting vehicle users within a region as nodes in the user intention influence topology graph, and abstracting the potential influence relationships between users based on load characteristic correlation as edges, the user intention influence topology graph can be constructed as follows: (10) in, N It is a set of nodes, consisting of car users within the region; F For node characteristics, it represents the set of user willingness to participate in scheduling; E It is a set of edges used to quantify the influence and transmission relationship of user nodes' intentions; A An adjacency matrix represents the connection relationships between nodes. A= ( a ij ), a ij These are the elements of the adjacency matrix.

[0045] Using the HSIC correlation coefficient between any two user nodes as the connectivity criterion, for any two user nodes i and nodes j compute nodes i and nodes j The HSIC correlation coefficient between them is used to determine the nodes. i and nodes j Are they connected? If the correlation coefficient is greater than the judgment threshold. Then it is considered a node. i and nodes j There must be an edge connecting them; otherwise, they are considered disconnected. The adjacency matrix is ​​determined by the following formula, and the weight of self-connected nodes in the matrix is ​​specifically set to 1.

[0046] (11) Adjacency Matrix A It can be represented as: (12) The process of transmitting vehicle user intentions based on the influence topology graph is as follows: Figure 5 As shown, the transmission process can be represented as: (13) in, To convey the user's intentions beforehand, To convey the user's wishes, Neighbor matrix A The degree matrix, This is the activation function.

[0047] To ensure that the five-level discrete structure is maintained after the user's intention is transmitted, this embodiment introduces an interval threshold mapping function after the intention transmission process to remap the intention value to the discrete level set {-2,-1,0,1,2}.

[0048] (14) in, To convey the user's wishes; This represents the mapped intention value; i 1 and i 2 is the mapping judgment threshold, which is set according to the distribution of user intentions.

[0049] Once the user's willingness transmission process is complete, their participation in demand response can be determined based on their final discrete willingness level. Users with a final willingness greater than 0 are considered willing to participate in scheduling; users with a final willingness less than or equal to 0 are considered not to participate in scheduling. Thus, the individual willingness of vehicle users within the region can be further aggregated into regional-level schedulable resources, providing user-side input for the subsequent MRIES three-layer game-theoretic scheduling model.

[0050] S30: Based on the final participation willingness level and operation status data, establish a three-level game scheduling model for a multi-regional integrated energy system. The three-level game scheduling model includes a master-slave game model between the dispatch center and the multi-regional integrated energy system, a cooperative game model between the integrated energy systems in each region, and a group game model based on the transmission of willingness among vehicle users in the region. The scheduling strategies of each subject are obtained by solving the three-level game scheduling model. The scheduling strategies include incentive price, power purchase, inter-regional interactive power, and equipment operation plan within each region.

[0051] In some embodiments, a three-tiered game theory architecture is established, consisting of a dispatch center, integrated energy systems in various regions, and vehicle user groups within the region, including: ① Master-slave game model between the scheduling center and MRIES The dispatch center, as the upper-level leader, is responsible for formulating demand response incentive prices and system dispatch strategies for multiple regions. Each RIES, as the operating entity, executes the dispatch center's decisions while coordinating the response behavior of the car user groups within its region. The car user groups within the region, as followers, have their demand response behavior jointly determined by the incentive price, individual characteristics, and the transmission of user intentions. The resulting incentive-based master-slave game model can be expressed as: (15) in, H It is an incentive-based master-slave game model; For the first participant set; This is the first set of strategies; JH This is the objective function of the model.

[0052] First Participant Set Represented as: (16) in, N DC Represented as the dispatch center; N RIESn For the first n RIES ( n =1,2,…); N EVn For the first n EVs ( n =1,2,…); N FCEVn For the first n FCEV users ( n =1,2,…).

[0053] First Strategy Set The decision variables for the game participants, including the incentive price, the power purchased by the upper-level buyers, and the response power of the user group, can be expressed as: (17) in, To incentivize prices; P grid The power purchased by the dispatch center from the upper-level power grid; P DR The schedulable power for actual user participation in demand response; P RIESn For the first n Each RIES purchases power from the dispatch center; The price at which DC power purchases energy from the upper-level power grid; The price at which each RIES purchases energy from the DC.

[0054] The goal of master-slave game theory is to minimize system operating costs and maximize self-profit, while fully considering the responses of lower-level user groups and the interaction power with each RIES. This can be represented as: (18) in, For system operating costs J buy For energy purchase costs; J inc To incentivize payment costs; This is for the energy interaction benefits between the dispatch center and each RIES.

[0055] ② Cooperative game model among the RIES Participants work together to achieve the optimal payoff through cooperative game theory. This method mainly shares operational information such as energy flow and price, without needing to exchange all private data within each region, thus possessing information independence and practical applicability.

[0056] The cooperative game model among the RIES is expressed as follows: (19) in, T A cooperative game model among the RIES; N T For the second participant set; For the second set of strategies; J T This is the objective function of the model.

[0057] Second Participant Set N T It can be represented as: (20) in, N RIESn For the first n RIES ( n =1,2,…).

[0058] Second strategy set The operational strategies for equipment within each region and the energy interaction decisions between regions consist of multiple continuous decision variables, including start-up and shutdown strategies, energy conversion output, energy storage and hydrogen storage operation, hydrogen energy system scheduling, transportation-side energy supply, and energy interaction between regions. This comprehensively describes the coordinated operation of RIES under the conditions of multi-energy coupling of electricity, heat, and hydrogen, and can be represented as: (twenty one) (twenty two) in, For the first i A set of strategies for each RIES; For unit start-up and shutdown strategies; The lateral interaction power between RIES; For the first i The interaction power between each RIES and the upstream DC; For the actual output of the photovoltaic unit; The actual output of the wind turbine; This refers to the actual output of the gas turbine; This refers to the actual output of the gas-fired boiler; This is the actual output of the waste heat recovery device; This represents the actual output of the heat exchange device; This refers to the actual output of the electrolytic cell unit; These are the energy storage / release capacities of energy storage devices (electric energy storage, thermal storage tanks, and hydrogen storage tanks); To actually contribute to the charging stations; To contribute actual power to the hydrogen refueling station; The energy purchase price between RIES and DC; The price of energy purchased between RIES.

[0059] The objective of the cooperative game among MRIES is to minimize the overall system operating cost by rationally formulating regional operating strategies based on energy prices and equipment operating status, while satisfying diverse user load demands. This can be expressed as: (twenty three) in, k Total number of energy types; n This represents the total number of RIES. For the first k The load demand for this type of energy; For the first k The unit price of energy sold for this type of energy; For operation and maintenance costs; For fuel costs.

[0060] ③ Group game model based on intention transmission among vehicle users within the region This embodiment constructs a graph-based user group response game model, where each user acts as a game participant. Through the constructed intentions influencing the topology graph, information exchange and strategy adjustments are made, forming a stable group response state without explicit negotiation, thereby achieving collaborative participation in demand response scheduling. The user game model constructed in this embodiment is applicable to both EV users and FCEV vehicle users; the main difference lies in the aggregation results of the group's schedulable power and hydrogen demand. The graph-based user game model can be represented as: (twenty four) in, G A user game model based on graph structure; N EV For the set of third participants; For the third set of strategies; J G This is the objective function of the model.

[0061] The third group of participants consisting of car users N EV , can be represented as: (25) in, N EVnFor the first n car users ( n =1,2,…).

[0062] Each car user's strategy in the game represents their willingness to participate in demand response; this is the third set of strategies. It can be represented as: (26) (27) in, For the first n The participation intention of individual car users is obtained by constructing a saturation incentive prediction model, which is used to characterize the individual response tendency of users without considering group influence. The fundamentally different values ​​are divided into five levels of participation intention: "very unwilling, unwilling, neutral, willing, and very willing." For the first i The status of individual car users' willingness to participate.

[0063] Considering user cooperation and information exchange, the core objective of a graph-based user cooperative game model is not to explicitly maximize the utility function of a single user, but rather to gradually reduce the differences in willingness within the group through mutual influence among users, thus forming a stable and consistent response state. This process can be implicitly represented by the following group consistency optimization objective: (28) in, For the first j The status of individual car users' willingness to participate.

[0064] S40: The game-reinforcement learning algorithm is used to solve the three-layer game scheduling model. The scheduling center is regarded as an intelligent agent, and the response behavior of the multi-region integrated energy system and its internal vehicle users is regarded as the environment. With the goal of minimizing the system operating cost, the optimal incentive price and regional scheduling decision are output.

[0065] Because the system exhibits significant nonlinearity, dynamic coupling, and non-analytical characteristics, traditional optimization methods are difficult to solve directly. This embodiment introduces a reinforcement learning method, modeling the incentive pricing and scheduling process of the scheduling center as an interactive learning process between an agent and a complex game environment. A game-reinforcement learning algorithm is proposed to achieve adaptive solution of the three-layer game scheduling model.

[0066] The energy dispatch center is treated as an intelligent agent, and the response behavior of MRIES and its internal vehicle population is considered as the environment, thus constructing a Markov decision process. Within each dispatch cycle, the agent issues incentive prices and dispatch decisions based on the current system state. The environment, based on these decisions, triggers the formation, transmission, and regional cooperative game processes of vehicle users' intentions, and returns the corresponding system operation results and reward signals. The game-reinforcement learning algorithm does not directly intervene in the charging and discharging decisions of individual vehicles, but rather indirectly influences the behavior of the user group by adjusting the macro-level signal of the incentive price.

[0067] The aggregated statistics output by the intention transmission mechanism are selected as the state input for the game-reinforcement learning algorithm. State vector S t Defined as: (29) (30) (31) (32) in, This represents the average saturation incentive price for the car group within the region, used to describe the overall price sensitivity of users, which is obtained by a weighted average of the saturation incentive levels for different vehicle types. V Indicates vehicle type; Indicates the area V Saturation excitation for vehicle class; This indicates the percentage weight of this type of vehicle within the region; The average willingness to participate in the group is obtained after the willingness is transmitted. P DR Scheduling power for user groups to participate in demand response; It represents the average energy state of the electric vehicle population and is used to constrain its continued participation capability; This is the system load state vector.

[0068] The actions of the reinforcement learning agent provide the energy dispatch center with real-time information. t The published demand response incentive prices and regional scheduling strategies are defined as the continuous action space: (33) in, π incn,t For the first n A regional integrated energy system t The incentive price decision variable at each moment. P RIESn For the first n Power dispatch decision variables for regional integrated energy systems.

[0069] In multi-region scenarios, the action vector is expanded into a set of differentiated incentive prices issued for different regions. The setting of the continuous action space enables the game-reinforcement learning algorithm to flexibly search for the optimal pricing strategy over a wider range.

[0070] Minimizing system operating costs is equivalently transformed into a reward maximization problem. A reward function is constructed by combining energy purchase costs, incentive payment costs, regional energy interaction benefits, and constraint violation penalties. Reward Function r t Defined as: (34) in, This represents the cost incurred by the energy dispatch center when purchasing energy from the higher-level energy network; This represents the incentive costs paid to car owners; This indicates the energy exchange benefits between the dispatch center and the integrated energy systems in various regions; This indicates a penalty introduced when power balance, constraints, or scheduling deviations are violated.

[0071] To further illustrate the feasibility and progressiveness of the method described above, simulation calculations are performed using the following method: The simulation targets three typical regional integrated energy systems: residential, commercial, and agricultural areas. The scheduling cycle is 24 hours, and the simulation step is 1 hour. The system connects two types of mobile energy storage users: electric vehicles (EVs) and hydrogen fuel cell vehicles (FCEVs). The total number of vehicles in the residential, commercial, and agricultural areas are 1000, 800, and 600, respectively, with EV / FCEV ratios of 70% / 30%, 50% / 50%, and 30% / 70%. For EVs, the battery capacity per vehicle is 60 kWh, the SOC operating range is 0.2–0.9, the target SOC is 0.8, and the maximum charging power and maximum discharging power per vehicle are 7 kW and 5 kW, respectively. For FCEVs, the hydrogen storage capacity per vehicle is 5 kg, the typical hydrogen refueling amount is 3 kg, the minimum remaining hydrogen storage is 0.8 kg, and the hydrogen consumption rate is 0.01 kg / km.

[0072] The effectiveness of the proposed method was verified using integrated energy system datasets from three typical regions: residential, commercial, and agricultural areas. The dataset includes integrated energy systems from these three regions, with a scheduling cycle of 24 hours and a simulation step size of 1 hour. Two types of mobile energy storage users were connected to the system: electric vehicles (EVs) and hydrogen fuel cell vehicles (FCEVs). The total number of vehicles in the residential, commercial, and agricultural areas were 1000, 800, and 600, respectively, with EV / FCEV ratios of 70% / 30%, 50% / 50%, and 30% / 70%. For EVs, the battery capacity per vehicle was 60 kWh, the SOC operating range was 0.2–0.9, the target SOC was 0.8, and the maximum charging and discharging power per vehicle was 7 kW and 5 kW, respectively. For FCEVs, the hydrogen storage capacity per vehicle was 5 kg, the typical hydrogen refueling amount was 3 kg, the minimum remaining hydrogen storage was 0.8 kg, and the hydrogen consumption rate was 0.01 kg / km. The system's electrical load, thermal load, and hydrogen load curves are as follows: Figure 6 As shown in Table 1, the time-of-use electricity price and time-of-use gas price are listed below.

[0073] Table 1. Time-of-use electricity and gas prices

[0074] To verify the effectiveness of the proposed MRIES collaborative optimization scheduling model, this embodiment sets up six different schemes for comparison.

[0075] Scheme 1: The DDPG dynamic incentive three-layer game collaborative optimization scheduling model proposed in this application is based on GCN intention transmission.

[0076] Option 2: A three-layer game-based collaborative optimization scheduling model with dynamic incentives based on initial intentions without considering intention propagation.

[0077] Option 3: A three-layer game-theoretic collaborative optimization scheduling model based on fixed incentives, considering the intention transmission of GCN.

[0078] Option 4: DDPG dynamic incentive independent optimization scheduling model based on GCN intention transmission.

[0079] Option 5: DDPG dynamic incentive independent optimization scheduling model based on initial intention without considering intention propagation.

[0080] Option 6: Consider a fixed-incentive independent optimization scheduling model based on GCN intention propagation.

[0081] The results of running schemes 1-6 are shown in Table 2.

[0082] Table 2 Comparison of running results

[0083] As shown in Table 2, Scheme 1 has the lowest overall cost, at only 125,600 yuan, which is significantly better than the other comparative schemes. This indicates that the DDPG dynamic incentive three-layer game collaborative optimization model based on GCN intention transmission proposed in this application has better economic efficiency.

[0084] The comparison between collaborative optimization and independent optimization shows that the comprehensive costs of schemes 1-3 are RMB 125,600, RMB 129,900, and RMB 151,900, respectively, all lower than the corresponding independent optimization schemes 4-6 (RMB 153,900, RMB 160,300, and RMB 164,900). This indicates that multi-regional collaborative scheduling can fully leverage the complementary advantages of each region in terms of load characteristics, flexible resources, and renewable energy output, thereby reducing the overall system operating cost.

[0085] From the perspective of the role of the user intention transmission mechanism, under the condition of collaborative optimization, the overall cost of Scheme 1, which considers GCN intention transmission, is 0.43 million yuan lower than that of Scheme 2, which does not consider intention transmission. This indicates that the constructed user intention transmission mechanism helps to improve the aggregation capability of demand response resources and enhance the supporting role of user-side adjustable resources in the optimized operation of the system.

[0086] From the perspective of incentive methods, Scheme 1 reduces the overall cost by 26,300 yuan compared to Scheme 3 with fixed incentives. This indicates that the DDPG dynamic incentive strategy can adaptively adjust the incentive level based on the system's operating status and user response, further improving the scheduling optimization effect. In summary, this invention has significant advantages in reducing system costs and improving collaborative operation capabilities.

[0087] The operation result diagram of the present invention is shown below. Figure 7 As shown, the overall electricity load in residential areas is relatively stable, with a slight increase from midday to evening. Demand is mainly met by wind and solar power output, grid purchases, energy storage discharge, and inter-regional power reception, reflecting the characteristics of a central power load. Heat load is mainly balanced by waste heat, gas-fired boilers, and thermal storage devices. The hydrogen system experiences relatively small fluctuations, primarily meeting basic energy needs, and its overall operation is relatively stable. Peak electricity load in commercial areas is concentrated during daytime business hours. Photovoltaic output and energy storage regulation have good timing matching, meeting local needs while also supplying electricity to other areas during certain periods. Heat load is highly persistent, with gas-fired boilers and thermal storage devices playing a significant role. The hydrogen system fluctuates significantly, reflecting frequent traffic activity in commercial areas and a strong periodic concentration of hydrogen demand. Agricultural areas demonstrate strong local energy supply and external support capabilities. After meeting local load demands, they can still supply electricity to other areas during multiple periods, making them the core power supply area in the system. Their heat load duration is relatively long, and heating demand is relatively stable. The hydrogen system exhibits significant hydrogen production, storage, and supply capabilities during certain periods, demonstrating a certain buffering and supplementary role for hydrogen energy.

[0088] For inter-regional power interaction, such as Figure 8As shown in the average inter-regional power exchange matrix, the agricultural area is the core power supply area in Scheme 1. The average power transmission from the residential area is close to zero, primarily acting as a power receiving area. The commercial area transmits an average of 10.15MW to the residential area, but no significant power to the agricultural area. The agricultural area, on the other hand, transmits an average of 145.81MW to the residential area and 67.02MW to the commercial area, significantly higher than the exchange levels between other areas. This indicates that the system does not exhibit disorderly and frequent bidirectional power exchanges, but rather forms a stable power flow structure with the agricultural area as the main supply node, the commercial area as an auxiliary support node, and the residential area as the main power receiving node. This demonstrates that coordinated optimization operation can establish a more economical and rational regional power supply and receiving relationship based on the load characteristics, energy output, and time-of-use operating status of each area, improving the overall economic efficiency and coordination of the system operation.

[0089] like Figure 9 and Figure 10 The figure shows the user participation rate distribution with and without intention-transfer. The participation rate trends for all user groups over 24 hours are generally consistent, with high participation rates concentrated in the previously more active time periods, and low participation rates primarily occurring near periods of weaker initial responses. This indicates that the intention-transfer mechanism has not altered the fundamental temporal patterns of user participation behavior. In comparison, Figure 9 The overall distribution of user participation rate is smoother, with some users showing increased participation rates at certain times, and local extreme differences have been mitigated to some extent. This indicates that the GCN intention transmission mechanism has enhanced the coordination and consistency of user group responses while retaining the original distribution characteristics.

[0090] Figure 11 This reflects the incremental changes in user participation rates before and after intention transmission. It can be seen that most increments fluctuate around 0, indicating that the correction of the original participation rate by intention transmission is generally gradual rather than abrupt. Meanwhile, some users show significant positive gains in specific periods, while a small number experience negative adjustments, suggesting that GCN propagation does not simply increase the overall participation rate, but rather structurally corrects user responses through neighborhood information interaction. Overall, this mechanism can guide group participation behavior towards a more stable and coordinated direction while preserving user heterogeneity, thereby improving the aggregation and schedulability of demand response resources.

[0091] This application also provides a three-layer game-theoretic optimization scheduling system for a multi-regional integrated energy system that considers the transmission of mobile energy storage intentions, such as... Figure 12 As shown, the three-layer game-theoretic optimization scheduling system for a multi-regional integrated energy system that considers the transmission of mobile energy storage intentions includes: The data acquisition module 1201 is configured to acquire the operation status data and vehicle user characteristic data of the multi-region integrated energy system. The operation status data includes the electricity, heat, and hydrogen load demand, new energy output, energy storage status, and the energy purchase price of the upper-level power grid in each region. The vehicle user characteristic data includes vehicle type, dispatchable capacity, real-time energy price, energy station congestion level, and time-series data of user load characteristics. The intention transmission module 1202 is configured to construct a vehicle user response intention transmission model based on the vehicle user characteristic data. The vehicle user response intention transmission model determines the initial response intention of individual users through saturation incentive prediction, constructs a user intention influence topology map by using load characteristic correlation analysis based on the Hilbert-Schmidt independence criterion, and iteratively transmits intention on the user intention influence topology map using a graph convolutional network to obtain the final participation intention level of the user group. The three-layer game module 1203 is configured to establish a three-layer game scheduling model for a multi-regional integrated energy system based on the final participation willingness level and the operating status data. The three-layer game scheduling model includes a master-slave game model between the scheduling center and the multi-regional integrated energy system, a cooperative game model between the integrated energy systems in each region, and a group game model based on willingness transmission among vehicle users in the region. The scheduling strategy of each subject is obtained by solving the three-layer game scheduling model. The scheduling strategy includes incentive price, power purchase capacity, inter-regional interactive power, and equipment operation plan within each region. The model solving module 1204 is configured to use a game-reinforcement learning algorithm to solve the three-layer game scheduling model, treating the scheduling center as an intelligent agent, the response behavior of the multi-region integrated energy system and its internal vehicle users as the environment, and outputting the optimal incentive price and regional scheduling decision with the goal of minimizing system operating costs.

[0092] This application provides an electronic device. The electronic device may include a processor and a memory, wherein the processor and the memory can communicate; exemplarily, the processor and the memory communicate via a communication bus.

[0093] The processor executes computer execution instructions stored in memory, causing the processor to perform the scheme in the above embodiments. The processor can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0094] The communication bus can be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus, etc. The system bus can be divided into address bus, data bus, control bus, etc. Transceivers are used to enable communication between the database access system and other computers (e.g., clients, read-write libraries, and read-only libraries). Memory may include random access memory (RAM) and may also include non-volatile memory.

[0095] The electronic device provided in this application embodiment can be the terminal device described in the above embodiments.

[0096] This application also provides a computer-readable storage medium storing computer instructions. When the computer instructions are executed on a computer, the computer performs the technical solution of the three-layer game optimization scheduling method for multi-region integrated energy systems that considers the transmission of mobile energy storage intentions, as described in the above embodiments.

[0097] This application also provides a computer program product, which includes a computer program stored in a computer-readable storage medium. At least one processor can read the computer program from the computer-readable storage medium. When the at least one processor executes the computer program, it can implement the technical solution of the three-layer game optimization scheduling method for multi-region integrated energy systems that considers the transmission of mobile energy storage intentions in the above embodiments.

[0098] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between systems or modules may be electrical, mechanical, or other forms.

[0099] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to implement the solution of this embodiment according to actual needs.

[0100] Furthermore, the functional modules in the various embodiments of this application can be integrated into one processing unit, or each module can exist physically separately, or two or more modules can be integrated into one unit. The unit composed of the above modules can be implemented in hardware or in the form of hardware plus software functional units.

[0101] The integrated modules described above, implemented as software functional modules, can be stored in a computer-readable storage medium. These software functional modules, stored in a storage medium, include several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute some steps of the methods of the various embodiments of this application.

[0102] It should be understood that the aforementioned processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. A general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this invention can be directly manifested as execution by a hardware processor, or execution by a combination of hardware and software modules within the processor.

[0103] The memory may include high-speed RAM, and may also include non-volatile storage (NVM), such as at least one disk storage device, and may also be a USB flash drive, external hard drive, read-only memory, disk or optical disc, etc.

[0104] Buses can be Industry Standard Architecture (ISA) buses, Peripheral Component Interconnect (PCI) buses, or Extended Industry Standard Architecture (EISA) buses, etc. Buses can be categorized into address buses, data buses, control buses, etc.

[0105] The aforementioned storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The storage medium can be any available medium that can be accessed by a general-purpose or special-purpose computer.

[0106] An exemplary storage medium is coupled to a processor, enabling the processor to read information from and write information to the storage medium. Alternatively, the storage medium can be an integral part of the processor. The processor and storage medium can reside in an Application Specific Integrated Circuit (ASIC). Alternatively, the processor and storage medium can exist as discrete components in an electronic control unit or main control device.

[0107] Those skilled in the art will understand that all or part of the steps of the above-described method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0108] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

Claims

1. A three-layer game-theoretic optimization scheduling method for a multi-regional integrated energy system considering the transmission of mobile energy storage intentions, characterized in that, The method includes: The system acquires operational status data and vehicle user characteristic data for a multi-regional integrated energy system. The operational status data includes electricity, heat, and hydrogen load demand, new energy output, energy storage status, and the energy purchase price of the upstream power grid in each region. The vehicle user characteristic data includes vehicle type, dispatchable capacity, real-time energy price, energy station congestion level, and time-series data of user load characteristics. Based on the vehicle user characteristic data, a vehicle user response intention transmission model is constructed. The vehicle user response intention transmission model determines the initial response intention of individual users through saturation incentive prediction. A load characteristic correlation analysis based on the Hilbert-Schmidt independence criterion is used to construct a user intention influence topology. A graph convolutional network is used to iteratively transmit intentions on the user intention influence topology to obtain the final participation intention level of the user group. Based on the final participation willingness level and the operational status data, a three-layer game scheduling model for a multi-regional integrated energy system is established. The three-layer game scheduling model includes a master-slave game model between the scheduling center and the multi-regional integrated energy system, a cooperative game model between the integrated energy systems in each region, and a group game model based on willingness transmission among vehicle users in the region. The scheduling strategies of each subject are obtained by solving the three-layer game scheduling model. The scheduling strategies include incentive price, power purchase capacity, inter-regional interactive power, and equipment operation plan within each region. The game-reinforcement learning algorithm is used to solve the three-layer game scheduling model. The scheduling center is regarded as an intelligent agent, and the response behavior of the multi-region integrated energy system and its internal vehicle users is regarded as the environment. With the goal of minimizing the system operating cost, the optimal incentive price and regional scheduling decision are output.

2. The method according to claim 1, characterized in that, The vehicle user response intention transmission model determines the initial response intention of an individual user through saturated incentive prediction in the following ways: The mapping relationship between the input feature variables and the saturation activation level, comprising five dimensions, is constructed as follows: （1） （2） in, Provide saturation incentives for vehicle users; The input feature vector; These are the trainable parameters of the model; V This indicates the vehicle type, used to distinguish between electric vehicles and hydrogen fuel cell vehicles; R Indicates the region type; The schedulable capacity represents the power difference between the current schedulable demand and the available capacity. This indicates the current energy price level. The congestion level of the energy station indicates its current utilization rate. f This is a mapping relationship; Based on the vehicle user saturation incentive, a probability model of vehicle user response intention is constructed, denoted as _____. The car at any time The probability of participation is The expression is: （3） in, For vehicle response incentive prices, It follows a Gaussian normal distribution; Map the participation probability to a discretized initial participation willingness level. , is represented as: （4） Among them, -2 means very unwilling, -1 means somewhat unwilling, 0 means neutral, 1 means somewhat willing, and 2 means very willing.

3. The method according to claim 1, characterized in that, A topology diagram of the impact of user intentions on load characteristics was constructed using load characteristic correlation analysis based on the Hilbert-Schmidt independence criterion, including: Based on the Hilbert-Schmidt independence criterion, the correlation coefficient between any two user load characteristics is calculated using the Hilbert-Schmidt independence criterion. The formula is as follows: （5） （6） （7） （8） in, For cross covariance operator The Hilbert-Schmidt norm; For bandwidth hyperparameters, x and x' Represents random variables X Two load characteristic samples; y and y′ Represents random variables Y Two load characteristic samples; K (x,x′) and L(y,y′) Defined in X and Y The Gaussian kernel function on the x-axis is used to characterize the nonlinear similarity between samples; exp(·) represents the exponential function. and These represent the kernel functions respectively. K and L The induced regenerative nucleus Hilbert space; E This represents the expectation operator, used to statistically average the kernel function values ​​among vehicle user load characteristic samples; HSIC The correlation coefficient is the Hilbert-Schmidt independence criterion. The load characteristics of each vehicle are selected as a sample, and the calculation is performed for each vehicle pair ( i,j The correlation coefficient between Hilbert-Schmidt independence criterion and the following is: （9） in, H ij To indicate vehicle user i With vehicle users j The correlation coefficient between the load characteristics; x i and x j They represent the first i The and the first j Load characteristics sample of individual vehicle users; The user intention influence topology graph is constructed by abstracting vehicle users within the region as nodes in the graph, and the potential influence relationships between users based on load characteristic correlation as edges. The graph is represented as follows: （10） in, N It is a set of nodes, consisting of car users within the region; F For node characteristics, it represents the set of user willingness to participate in scheduling; E It is a set of edges used to quantify the influence and transmission relationship of user nodes' intentions; A An adjacency matrix represents the connection relationships between nodes. A= ( a ij ), a ij These are elements of the adjacency matrix; Using the HSIC correlation coefficient as the connectivity criterion, for any two user nodes i and nodes j If the correlation coefficient is greater than the judgment threshold If a node is connected to another node, it is considered to have an edge connection; otherwise, it is not connected. The adjacency matrix is ​​determined by the following formula, and the weight of the self-connected node is set to 1: （11） Adjacency matrix A Represented as: （12） in, a n,i For vehicle users n With vehicle users i Adjacency matrix elements between them a i,n For vehicle users i With vehicle users n The adjacent matrix elements between them.

4. The method according to claim 3, characterized in that, By using a graph convolutional network to iteratively propagate user intentions across the topology graph influencing user intentions, the final participation intention level of the user group is obtained, including: The process of iteratively transmitting user intentions on the topology graph that influences user intentions using a graph convolutional network is represented as follows: （13） in, To convey the user's intentions beforehand, To convey the user's wishes, Neighbor matrix A The degree matrix, For activation functions; After the intention transmission process, an interval threshold mapping function is introduced to remap the intention values ​​to the discrete level set {-2,-1,0,1,2}: (14) in, To convey the user's wishes; This represents the mapped intention value; i 1 and i 2 represents the mapping judgment threshold, which is set based on the distribution of user intentions; Once the user's intention transmission process is complete, their participation in demand response is determined based on their final discrete intention level: users with a final intention greater than 0 are considered to be willing to participate in scheduling; users with a final intention less than or equal to 0 are considered not to participate in scheduling.

5. The method according to claim 1, characterized in that, The master-slave game model between the dispatch center and the multi-regional integrated energy system is expressed as follows: (15) in, H It is an incentive-based master-slave game model; For the first participant set; This is the first set of strategies; J H The objective function of the model; First Participant Set Represented as: (16) in, N DC Represented as the dispatch center; N RIESn For the first n A regional integrated energy system; N EVn For the first n One electric vehicle user; N FCEVn For the first n One hydrogen fuel cell vehicle user; First Strategy Set Represented as: （17） in, To incentivize prices; P grid The power purchased by the dispatch center from the upper-level power grid; P DR The schedulable power for actual user participation in demand response; P RIESn For the first n Each RIES purchases power from the dispatch center; The price at which DC power purchases energy from the upper-level power grid; The price at which each RIES purchases energy from the DC; The objective of a master-slave game is expressed as: （18） in, For system operating costs J buy For energy purchase costs; J inc To incentivize payment costs; For the energy interaction benefits between the dispatch center and each RIES; The cooperative game model among the integrated energy systems in different regions is expressed as follows: （19） in, T A cooperative game model among the RIES; N T For the second participant set; For the second set of strategies; J T The objective function of the model; Second Participant Set N T Represented as: （20） in, N RIESn For the first n RIES; Second strategy set Represented as: （21） （22） in, For the first i A set of strategies for each RIES; For unit start-up and shutdown strategies; The lateral interaction power between RIES; For the first i The interaction power between each RIES and the upstream DC; For the actual output of the photovoltaic unit; The actual output of the wind turbine; This refers to the actual output of the gas turbine; This refers to the actual output of the gas-fired boiler; This is the actual output of the waste heat recovery device; This represents the actual output of the heat exchange device; This refers to the actual output of the electrolytic cell unit; These refer to the energy storage / release capacity of the energy storage device; To actually contribute to the charging stations; To contribute actual power to the hydrogen refueling station; The energy purchase price between RIES and DC; For RIES inter-energy purchase price; The objective function of cooperative game is: （23） in, k Total number of energy types; n This represents the total number of RIES. For the first k The load demand for this type of energy; For the first k The unit price of energy sold for this type of energy; For operation and maintenance costs; For fuel costs.

6. The method according to claim 1, characterized in that, The group game model based on intention transmission among vehicle users in the area is represented as follows: （24） in, G A user game model based on graph structure; N EV For the set of third participants; For the third set of strategies; J G The objective function of the model; The third group of participants consisting of car users N EV Represented as: （25） in, N EVn For the first n One car user; Each car user's strategy in the game represents their willingness to participate in demand response; this is the third set of strategies. Represented as: （26） （27） in, For the first n The willingness of individual car users to participate; For the first i The participation status of individual car users; The optimization objective of the group game is a group consistency optimization objective: （28） in, For the first j The status of individual car users' willingness to participate. The group game model is solved implicitly through a willingness transmission mechanism based on graph convolutional networks, iteratively updating user policies until convergence. The final willingness value after convergence is used to determine whether users participate in demand response and to calculate the group schedulable power.

7. The method according to claim 1, characterized in that, The process of solving the three-layer game scheduling model using the game-reinforcement learning algorithm includes: Treating the dispatch center as an intelligent agent and the response behavior of the multi-regional integrated energy system and its internal vehicle users as the environment, a Markov decision process is constructed, in which: State vector S t Defined as: （29） （30） （31） （32） in, This represents the average saturation incentive price for the car group within the region, used to describe the overall price sensitivity of users, which is obtained by a weighted average of the saturation incentive levels for different vehicle types. V Indicates vehicle type; Indicates the area V Saturation excitation for vehicle class; This indicates the percentage weight of this type of vehicle within the region; The average willingness to participate in the group is obtained after the willingness is transmitted. P DR Scheduling power for user groups to participate in demand response; It represents the average energy state of the electric vehicle population and is used to constrain its continued participation capability; This is the system load state vector; Actions of the intelligent agent a t Represented as: （33） in, π incn,t For the first n A regional integrated energy system t The incentive price decision variable at each moment. P RIESn For the first n Power dispatch decision variables for regional integrated energy systems. reward function r t Defined as: （34） in, This represents the cost incurred by the energy dispatch center when purchasing energy from the higher-level energy network; This represents the incentive costs paid to car owners; This indicates the energy exchange benefits between the dispatch center and the integrated energy systems in various regions; This indicates a penalty term introduced when power balance, constraints, or scheduling deviations are violated; The agent is based on the current state S t Output Action a t and according to the reward function r t The strategy is updated, and the final output is the optimal incentive price and regional scheduling decision that maximizes the cumulative reward.

8. A three-layer game-theoretic optimization scheduling system for a multi-regional integrated energy system considering the transmission of mobile energy storage intentions, characterized in that, The system includes: The data acquisition module is configured to acquire the operation status data and vehicle user characteristic data of the multi-regional integrated energy system. The operation status data includes the electricity, heat, and hydrogen load demand, new energy output, energy storage status, and the energy purchase price of the upper-level power grid in each region. The vehicle user characteristic data includes vehicle type, dispatchable capacity, real-time energy price, energy station congestion level, and time-series data of user load characteristics. The intention transmission module is configured to construct a vehicle user response intention transmission model based on the vehicle user characteristic data. The vehicle user response intention transmission model determines the initial response intention of individual users through saturation incentive prediction, constructs a user intention influence topology map by using load characteristic correlation analysis based on the Hilbert-Schmidt independence criterion, and uses a graph convolutional network to iteratively transmit intentions on the user intention influence topology map to obtain the final participation intention level of the user group. The three-layer game module is configured to establish a three-layer game scheduling model for a multi-regional integrated energy system based on the final participation willingness level and the operating status data. The three-layer game scheduling model includes a master-slave game model between the scheduling center and the multi-regional integrated energy system, a cooperative game model between the integrated energy systems in each region, and a group game model based on willingness transmission among vehicle users in the region. The scheduling strategy of each subject is obtained by solving the three-layer game scheduling model. The scheduling strategy includes incentive price, power purchase capacity, inter-regional interactive power, and equipment operation plan within each region. The model solving module is configured to use a game-reinforcement learning algorithm to solve the three-layer game scheduling model. The scheduling center is regarded as an intelligent agent, and the response behavior of the multi-region integrated energy system and its internal vehicle users is regarded as the environment. With the goal of minimizing the system operating cost, the module outputs the optimal incentive price and regional scheduling decision.

9. An electronic device, characterized in that, include: A processor, and a memory communicatively connected to the processor; The memory stores computer-executed instructions; The processor executes the computer execution instructions stored in the memory to implement the three-layer game optimization scheduling method for multi-region integrated energy systems that considers the transmission of mobile energy storage intentions as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the three-layer game-theoretic optimization scheduling method for multi-regional integrated energy systems that considers the transmission of mobile energy storage intentions as described in any one of claims 1-7.

Images