Multi-microgrid multivariate scheduling method, system and device based on virtual energy storage and medium

Abstract

The invention relates to the technical field of multi-microgrid management, and discloses a multi-microgrid multivariate scheduling method, system, device and medium based on virtual energy storage, and the method comprises the steps: inputting all kinds of determined virtual energy storage real-time operation data into a target Markov game model, and carrying out the scheduling of a target multi-microgrid according to an obtained scheduling strategy, the construction of the target Markov game model comprises the following steps: constructing an operation benefit model and an operation cost model of the micro-grid according to the operation data analysis result of each type of virtual energy storage, and further constructing a micro-grid optimization model of each micro-grid and a multi-micro-grid optimization model; and carrying out local model training on an initial Markov game model obtained according to the micro-grid optimization model and the multi-micro-grid optimization model, and carrying out iterative updating on the obtained noise gradient of each intelligent agent by adopting a federated learning algorithm to obtain a target Markov game model. According to the method provided by the invention, unified multivariate energy scheduling is carried out on multiple microgrids, and the data security is ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of multi-microgrid management, and in particular to a multi-microgrid multi-element scheduling method, system, device and medium based on virtual energy storage. Background Art

[0002] Against the backdrop of energy transition, the integration of large-scale, highly volatile renewable energy into the grid has become a key trend in power system development. However, the intermittent and random nature of this type of renewable energy poses significant challenges to the stable operation of power systems. Multi-microgrids, consisting of interconnected microgrids, offer an effective solution to these challenges due to their flexible operation and collaborative capabilities. Through energy exchange and coordinated operation among multiple microgrids, multi-microgrids can, to a certain extent, mitigate the fluctuations of renewable energy and enhance the stability and reliability of power systems.

[0003] In order to achieve energy interaction and coordinated operation among multiple microgrids, it is necessary to coordinate the source and load scheduling of multiple microgrids. However, the source and load resource endowments of each microgrid in the current multi-microgrid are different. The configuration proportion and capacity scale of distributed renewable energy on the power supply side vary significantly. The spatiotemporal characteristics and adjustable potential of various distributed controllable loads on the user side are different. The existing technology lacks a unified multi-dimensional optimization scheduling strategy, and there is a lack of effective protection for the various types of equipment data and user-side data collected during the scheduling process, which poses a risk of data leakage.

[0004] It can be seen that how to achieve multi-dimensional optimized scheduling of multiple microgrids and improve the data security of multiple microgrids has become a technical problem that needs to be solved urgently by those skilled in the art. Summary of the Invention

[0005] The present invention provides a multi-microgrid multi-scheduling method, system, device and medium based on virtual energy storage to solve the technical problems of how to achieve multi-optimal scheduling of multi-microgrids and improve the data security of multi-microgrids, thereby achieving multi-optimal scheduling of multi-microgrids and improving the data security of multi-microgrids.

[0006] In a first aspect, the present invention provides a multi-microgrid multi-element scheduling method based on virtual energy storage. The method comprises: determining the virtual energy storage type of a target multi-microgrid, inputting first real-time operating data of each type of virtual energy storage and second real-time operating data of the target multi-microgrid into a target Markov game model, and scheduling the target multi-microgrid according to the obtained scheduling strategy. The construction and training process of the target Markov game model comprises: Constructing an operation model of each microgrid in the target multi-microgrid according to the analysis results of the operation data of each type of virtual energy storage, wherein the operation model includes an operation benefit model and an operation cost model; Constructing a microgrid optimization model for each microgrid according to the output results corresponding to the operation benefit model and the operation cost model, and optimizing the target multi-microgrid according to the microgrid optimization model of each microgrid to obtain a multi-microgrid optimization model; Constructing an initial Markov game model based on the microgrid optimization model and the multi-microgrid optimization model, and performing local model training on the initial Markov game model to obtain a noise gradient for each intelligent agent, wherein the initial Markov game model defines the intelligent agent based on each microgrid, and converting the microgrid optimization model into the local model and converting the multi-microgrid optimization model into a global model; A federated learning algorithm is used to update the noise gradient of each agent, and the local model is iteratively trained according to the obtained global model noise gradient until the local model meets the preset conditions, thereby obtaining the target Markov game model.

[0007] Preferably, the step of constructing an operation model of each microgrid in the target multi-microgrid based on the analysis results of the operation data of each type of virtual energy storage includes: Based on the switching characteristics of electric vehicles under the guidance of time-of-use electricity prices, electric vehicle operation data is analyzed to construct an electric vehicle virtual energy storage relationship. The electric vehicle virtual energy storage relationship includes: an electric vehicle virtual capacitance value relationship, an electric vehicle virtual state of charge relationship, and an electric vehicle battery aging relationship. The process of controlling the rotor kinetic energy by adjusting the motor state of the wind turbine is equivalent to the charging and discharging process of the wind turbine, and the wind turbine operating data is analyzed to construct a wind turbine virtual energy storage relationship, which includes: a wind turbine virtual capacitance value relationship and a wind turbine virtual state of charge relationship; The response process of the curtailable load, transferable load, and shiftable load of the controllable load is equated to the charging and discharging process of the controllable load, and the controllable load operation data is analyzed to construct a virtual energy storage relationship of the controllable load, wherein the virtual energy storage relationship of the controllable load includes: a virtual power value relationship of the curtailable load, a virtual power value relationship of the transferable load, and a virtual power value relationship of the shiftable load; According to the energy conversion process of the hydrogen storage device, the operating data of the hydrogen storage device is analyzed to construct a virtual energy storage relationship of the hydrogen storage device, which includes: a relationship between the amount of hydrogen produced by the alkaline electrolyzer, a relationship between the amount of hydrogen burned by the proton exchange membrane fuel cell, a relationship between the virtual state of charge of the hydrogen storage tank, and a relationship between the virtual capacitance value of the hydrogen storage device; Constructing an operation benefit model for each microgrid in the target multi-microgrid according to the electric vehicle virtual energy storage relationship and the hydrogen storage device virtual energy storage relationship; An operation cost model of each microgrid in the target multi-microgrid is constructed based on the electric vehicle virtual energy storage relationship formula, the wind turbine virtual energy storage relationship formula, the controllable load virtual energy storage relationship formula and the hydrogen storage device virtual energy storage relationship formula.

[0008] Preferably, the operation benefit model of each microgrid in the target multi-microgrid is constructed based on the electric vehicle virtual energy storage relationship and the hydrogen storage device virtual energy storage relationship, including: Obtaining a first data set of electric vehicle virtual energy storage participating in charging scheduling for each microgrid within the target multi-microgrid according to the electric vehicle virtual energy storage relationship formula, and obtaining a charging and discharging benefit relationship formula for each microgrid within the target multi-microgrid based on the first data set, wherein the first data set includes at least: a total number of first electric vehicle virtual energy storage participating in charging scheduling and a charging amount of the electric vehicle virtual energy storage; Obtaining a second data set of virtual energy storage of the hydrogen storage device of each microgrid according to the virtual energy storage relationship of the hydrogen storage device, and obtaining a hydrogen production benefit relationship of each microgrid based on the second data set, wherein the second data set includes at least: daily hydrogen quality and hydrogen production of the alkaline electrolyzer; Obtaining the power sold by each microgrid to other microgrids, and obtaining a power sales benefit relationship formula for each microgrid based on the power sold; An operation benefit model of each microgrid in the target multi-microgrid is constructed based on the electric vehicle charging and discharging benefit relationship, the hydrogen production benefit relationship, and the electricity sales benefit relationship.

[0009] Preferably, the operation cost model of each microgrid in the target multi-microgrid is constructed according to the electric vehicle virtual energy storage relationship, the wind turbine virtual energy storage relationship, the controllable load virtual energy storage relationship, and the hydrogen storage device virtual energy storage relationship, including: Obtaining a third data set of electric vehicle virtual energy storage participating in discharge scheduling for each microgrid within the target multi-microgrid based on the electric vehicle virtual energy storage relationship formula, and obtaining a scheduling cost relationship formula for the electric vehicle virtual energy storage for each microgrid within the target multi-microgrid based on the third data set, wherein the third data set includes at least: a total number of second electric vehicle virtual energy storage participating in discharge scheduling and a discharge amount of the electric vehicle virtual energy storage; Obtaining a fourth data group of wind turbine virtual energy storage wind regulation of each microgrid according to the wind turbine virtual energy storage relationship formula, and obtaining a wind turbine virtual energy storage scheduling cost relationship formula for each microgrid based on the fourth data group, wherein the fourth data group includes at least: the total number of wind turbine virtual energy storage, the output power of the wind turbine virtual energy storage before wind regulation, and the output power of the wind turbine virtual energy storage after wind regulation; Obtaining a fifth data group for the controllable load virtual energy storage of each microgrid participating in peak regulation based on the controllable load virtual energy storage relationship formula, and obtaining a controllable load virtual energy storage scheduling cost relationship formula for each microgrid based on the fifth data group, wherein the fifth data group includes at least: a power value of a curtailable load that can participate in scheduling, a power value of a transferable load that can participate in scheduling, and a power value of a shiftable load that can participate in scheduling; Obtaining a sixth data set of the virtual energy storage participation of the hydrogen storage device of each microgrid in peak load regulation according to the hydrogen storage device virtual energy storage relationship formula, and obtaining a virtual energy storage scheduling cost relationship formula of the hydrogen storage device of each microgrid based on the sixth data set, wherein the sixth data includes at least: the power of the alkaline electrolyzer, the operating status of the alkaline electrolyzer, the power of the proton exchange membrane fuel cell, and the operating status of the proton exchange membrane fuel cell; Obtaining other operating cost relationship formulas for each of the microgrids, wherein the other operating costs include at least: a microgrid network loss cost relationship formula, a battery cost relationship formula, a photovoltaic operation and maintenance cost relationship formula, and a microgrid electricity purchase cost relationship formula; According to the electric vehicle virtual energy storage scheduling cost relationship formula, the wind turbine virtual energy storage scheduling cost relationship formula, the controllable load virtual energy storage scheduling cost relationship formula, the hydrogen storage device virtual energy storage scheduling cost relationship formula and the other operating cost relationship formulas, an operating cost model of each microgrid in the target multi-microgrid is constructed.

[0010] Preferably, constructing an initial Markov game model based on the microgrid optimization model and the multi-microgrid optimization model, and performing local model training on the initial Markov game model to obtain the noise gradient of each intelligent agent includes: Defining each microgrid in the target multi-microgrid as an intelligent agent, obtaining a state space for observing each microgrid, wherein the state space includes an overall state space for observing the microgrid and a local state space for observing the microgrid; Obtaining an action space of each of the intelligent agents, the action space including at least: a first charge and discharge power of each type of virtual energy storage, a second charge and discharge power of the battery, and a purchase and sale power of the microgrid; Constructing a reward function for the target multi-microgrid according to the multi-microgrid optimization model, wherein the reward function includes: an optimization goal achievement reward and a constraint violation penalty reward; Converting the microgrid optimization model into a local model and converting the multi-microgrid optimization model into a global model; constructing an initial Markov game model according to the state space, the action space, the reward function, the local model, and the global model; A deep deterministic policy gradient reinforcement learning method is used to perform local model training on the initial Markov game model to obtain the latest gradient of each local model, and Laplace noise is introduced into the latest gradient to obtain the noise gradient of each intelligent agent.

[0011] Preferably, the reward function is expressed as: in, represents the reward function, represents the weight coefficient of the optimization objective of the multi-microgrid optimization model, represents the optimization objective function of the multi-microgrid optimization model, represents the weight coefficient of the constraint violation penalty, Represents the penalty cost of exceeding the limit for various constraints in the multi-microgrid optimization model.

[0012] Preferably, the method of updating the noise gradient of each agent using a federated learning algorithm and iteratively training the local model according to the obtained global model noise gradient until the local model meets preset conditions to obtain a target Markov game model includes: Uploading the noise gradient of each agent to a global server to update global model parameters to obtain a global model noise gradient; The global model noise gradient is globally broadcasted to update the noise gradient of each of the intelligent agents, and iterative training is performed based on the updated noise gradient until the local model meets preset conditions to obtain the target Markov game model. The preset conditions include: the local model loss curve converges and / or the local model gain curve converges.

[0013] In a second aspect, the present invention further provides a multi-microgrid multi-dispatching system based on virtual energy storage, which implements the multi-microgrid multi-dispatching method based on virtual energy storage described above, and the system includes: a multi-dispatching module; The multivariate scheduling module is used to determine the virtual energy storage type of the target multi-microgrid, input the first real-time operation data of each type of virtual energy storage and the second real-time operation data of the target multi-microgrid into the target Markov game model, and schedule the target multi-microgrid according to the obtained scheduling strategy; The multivariate dispatch module includes: a microgrid operation model construction unit, a multi-microgrid optimization model construction unit, an initial Markov game model construction unit and a model iteration training unit; The microgrid operation model construction unit is used to construct an operation model of each microgrid in the target multi-microgrid according to the operation data analysis results of each type of virtual energy storage, wherein the operation model includes an operation benefit model and an operation cost model; The multi-microgrid optimization model construction unit is configured to construct a microgrid optimization model for each microgrid according to the output results of the operation benefit model and the operation cost model, and optimize the target multi-microgrid according to the microgrid optimization model of each microgrid to obtain a multi-microgrid optimization model; The initial Markov game model construction unit is configured to construct an initial Markov game model based on the microgrid optimization model and the multi-microgrid optimization model, and perform local model training on the initial Markov game model to obtain a noise gradient of each intelligent agent. The initial Markov game model defines the intelligent agent based on each microgrid, and converts the microgrid optimization model into the local model and converts the multi-microgrid optimization model into a global model. The model iterative training unit is used to update the noise gradient of each intelligent agent using a federated learning algorithm, and iteratively train the local model according to the obtained global model noise gradient until the local model meets the preset conditions, thereby obtaining the target Markov game model.

[0014] In a third aspect, the present invention further provides a computer device comprising a memory, a processor, and a transceiver, which are connected via a bus; the memory is used to store a set of computer program instructions and data, and transmit the stored data to the processor, and the processor executes the computer program instructions stored in the memory to execute the above-mentioned multi-microgrid multi-element scheduling method based on virtual energy storage.

[0015] In a fourth aspect, the present invention further provides a computer-readable storage medium, wherein a computer program is stored in the computer-readable storage medium. When the computer program is executed, the multi-microgrid multi-dispatching method based on virtual energy storage described above is implemented.

[0016] This application provides a multi-microgrid multi-element scheduling method, system, device, and medium based on virtual energy storage. Compared with the existing technology, the embodiments of this application have the following beneficial effects: The multi-microgrid multi-dispatching method based on virtual energy storage disclosed in this application virtualizes electric vehicles, wind turbines, controllable loads and hydrogen storage devices into the form of energy storage, uses virtual capacitance values ​​as standardized energy control parameters, simplifies the operating parameters of the flexible and adjustable resources of the multi-microgrid, and thus improves the perception level of the energy regulation potential of the multi-microgrid. Using the virtual state of charge as the state criterion for virtual energy storage, it provides a unified quantitative benchmark for the evaluation of the frequency and peak regulation capabilities of the multi-microgrid. The Markov game algorithm fully utilizes the synergistic potential of the source, grid, load and storage of the multi-microgrid, perceives the distributed renewable energy configuration on the power supply side of each microgrid and the various distributed controllable loads on the user side, and realizes unified multi-energy scheduling of the multi-microgrid, meets the needs of multi-microgrid coordinated control, significantly improves the benefits of coordinated optimization, realizes the rational allocation of resources, improves the utilization efficiency of multi-microgrid resources, and ensures the data security of each microgrid. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 This is a schematic diagram of the steps of the training process for constructing a target Markov game model provided by a preferred embodiment of the present invention; Figure 2 This is a structural diagram of a multi-microgrid multi-element dispatching system based on virtual energy storage provided by a preferred embodiment of the present invention; Figure 3 is an internal structural diagram of a computer device according to an embodiment of the present invention; Reference numerals: 1-Microgrid operation model construction unit, 2-Multi-microgrid optimization model construction unit, 3-Initial Markov game model construction unit, 4-Model iterative training unit. DETAILED DESCRIPTION

[0018] The following is a detailed explanation of the embodiments of the present invention in conjunction with the accompanying drawings. The embodiments are provided for illustrative purposes only and cannot be understood as limitations on the present invention. The accompanying drawings are for reference and illustration purposes only and do not constitute a limitation on the scope of patent protection of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. In the description of the present invention, the terms "first", "second", "third", etc. are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first", "second", "third", etc. may explicitly or implicitly include one or more of the features. In the description of the present invention, unless otherwise specified, the meaning of "multiple" is two or more.

[0019] In the description of the present invention, it should be noted that, unless otherwise expressly specified and limited, the terms "installed", "connected" and "connected" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or an indirect connection through an intermediate medium, or it can be a communication between the two components. The terms "vertical", "horizontal", "left", "right", "up", "down" and similar expressions used herein are for illustrative purposes only, and do not indicate or imply that the device or component referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention. The term "and / or" used herein includes any and all combinations of one or more related listed items. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0020] In describing the present invention, it should be noted that, unless otherwise defined, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art. The terms used in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. Those skilled in the art will understand the specific meanings of the above terms in the present invention in specific circumstances.

[0021] In an embodiment of the present invention, a multi-microgrid multi-element scheduling method based on virtual energy storage is provided, the method comprising: Determine the virtual energy storage type of the target multi-microgrid, and input the first real-time operation data of each type of virtual energy storage and the second real-time operation data of the target multi-microgrid into the target Markov game model, and schedule the target multi-microgrid according to the obtained scheduling strategy, such as Figure 1 The figure shows the steps of the training process for building the target Markov game model, including: S1. Based on the analysis results of the operational data of each type of virtual energy storage, an operational model for each microgrid within the target multi-microgrid is constructed. The operational model includes an operational benefit model and an operational cost model. In a preferred embodiment of the present application, virtual energy storage is introduced. Virtual energy storage is a new energy storage concept that integrates and coordinates multiple decentralized, adjustable resources to simulate the functions of traditional physical energy storage. Depending on the operational characteristics of the microgrid, virtual energy storage includes at least: electric vehicle virtual energy storage, wind turbine virtual energy storage, controllable load virtual energy storage, and hydrogen storage device virtual energy storage.

[0022] According to the characteristics of electric vehicles being chargeable and dischargeable, electric vehicles are virtualized as virtual energy storage. Based on the switching characteristics of electric vehicles under the guidance of time-of-use electricity prices, electric vehicle operation data is analyzed to construct an electric vehicle virtual energy storage relationship. The electric vehicle virtual energy storage relationship includes: the electric vehicle virtual capacitance value relationship, the electric vehicle virtual state of charge relationship, and the electric vehicle battery aging relationship. The electric vehicle virtual energy storage relationship is expressed as follows: in, Indicates the time period, Indicates the total time period, express The virtual capacitance value of electric vehicles during this period, express The charging and discharging active power of electric vehicles during the period, express The DC voltage of the charging station during the period, express The virtual state of charge of electric vehicles during the period, express The virtual state of charge of electric vehicles during the period, Indicates the charge and discharge loss coefficient of electric vehicle batteries, Indicates the battery capacity of electric vehicles, express The aging degree of electric vehicle batteries during this period, express The aging degree of electric vehicle batteries during this period, represents the attenuation coefficient of the electric vehicle battery, Indicated by Time period to The aging capacity of electric vehicle batteries during the period.

[0023] An electric vehicle virtual energy storage model is constructed according to the electric vehicle virtual energy storage relationship to model and control the electric vehicle virtual energy storage.

[0024] Since the kinetic energy of the fan is affected by the rotor speed, the process of controlling the rotor kinetic energy by adjusting the fan motor state is equivalent to the charging and discharging process of the fan. The fan operation data is analyzed to construct the fan virtual energy storage relationship. The fan virtual energy storage relationship includes the fan virtual capacitance value relationship and the fan virtual charge state relationship. The fan virtual energy storage relationship is expressed as follows: in, express The virtual capacitance value of the fan during the period, Indicates the maximum power point tracking coefficient of the wind turbine, Indicates the rated value of the fan rotor angular velocity, express The fan rotor angular velocity during this period, Indicates the voltage offset at the fan terminal. express Virtual state of charge of wind turbine during the period.

[0025] A wind turbine virtual energy storage model is constructed according to the wind turbine virtual energy storage relationship to model and control the wind turbine virtual energy storage.

[0026] According to the adjustability of controllable loads, the loads participating in demand response can be divided into curtailable loads, transferable loads, and shiftable loads. The response processes of curtailable loads, transferable loads, and shiftable loads are equivalent to the charging and discharging processes of controllable loads. The operating data of controllable loads are analyzed to construct a virtual energy storage relationship for controllable loads. The virtual energy storage relationship for controllable loads includes the virtual power value relationship for curtailable loads, the virtual power value relationship for transferable loads, and the virtual power value relationship for shiftable loads. The virtual energy storage relationship for controllable loads is expressed as follows: in, express The virtual power value of the load can be reduced during the period. express The power before the load participates in the dispatch can be reduced during the period. express The load power that can be reduced during the scheduling period is: express The virtual power value of the load that can be transferred during the time period, express The power of the load that can be transferred before it participates in the dispatching, express The transferable load power participating in the dispatch during the time period, 、 and represents a 0-1 variable, express The power of the load increase can be transferred during the period. express The power of load reduction can be transferred during the period. express The virtual power value of the load can be shifted in time period. express The power before the load can be shifted during the period to participate in the dispatch, express The load power that can be shifted during the scheduling period, express The power of the load increase can be shifted during the period. express The time period can shift the power of load reduction.

[0027] A controllable load virtual energy storage model is constructed according to the controllable load virtual energy storage relationship to perform modeled control on the controllable load virtual energy storage.

[0028] Based on the electricity-hydrogen-electricity energy conversion process of the alkaline electrolyzer using electricity to produce hydrogen, the hydrogen storage tank compressing hydrogen storage, and the proton exchange membrane fuel cell hydrogen combustion output power in the hydrogen storage device, and analyzing the hydrogen storage device operation data, a virtual energy storage relationship for the hydrogen storage device is constructed. The virtual energy storage relationship for the hydrogen storage device includes: the relationship between the hydrogen production amount of the alkaline electrolyzer, the relationship between the hydrogen combustion amount of the proton exchange membrane fuel cell, the relationship between the virtual state of charge of the hydrogen storage tank, and the relationship between the virtual capacitance value of the hydrogen storage device. The virtual energy storage relationship for the hydrogen storage device is expressed as follows: in, express The amount of hydrogen produced by alkaline electrolyzer during the period, Indicates the conversion efficiency of the alkaline electrolyzer, express The power of the alkaline electrolyzer during the period, represents the conversion efficiency of the alkaline electrolyzer converter, Indicates the operation status of the alkaline electrolyzer, Indicates the lower heating value of hydrogen, express The amount of hydrogen burned in the proton exchange membrane fuel cell during this period, express The power of the proton exchange membrane fuel cell during the period, Indicates the operation status of the proton exchange membrane fuel cell, represents the power generation efficiency of the proton exchange membrane fuel cell, represents the efficiency of the converter, express Virtual state of charge of hydrogen storage tank during the period, express Virtual state of charge of hydrogen storage tank during the period, Indicates the maximum hydrogen storage mass of the hydrogen storage tank, express The voltage of the hydrogen storage device during the period.

[0029] A virtual energy storage model of the hydrogen storage device is constructed according to the virtual energy storage relationship of the hydrogen storage device to perform modeling control on the virtual energy storage of the hydrogen storage device.

[0030] Furthermore, a microgrid operating benefit model is constructed based on the virtual energy storage equations for electric vehicles and hydrogen storage devices. The peak-shaving benefits of wind turbine virtual energy storage and controllable load virtual energy storage fall within the scope of the electricity market. Therefore, in the preferred embodiment of this application, for each microgrid, the benefits of participating in scheduling for both the virtual energy storage of electric vehicles and hydrogen storage devices are considered to construct equations for the benefits of electric vehicle charging and discharging, hydrogen production, and electricity sales.

[0031] Specifically, a first data set of electric vehicle virtual energy storage participating in charging scheduling of each microgrid in the target multi-microgrid is obtained based on the electric vehicle virtual energy storage relationship. The first data set includes at least: the total number of first electric vehicle virtual energy storage participating in charging scheduling and the amount of electric vehicle virtual energy storage charging. Based on the first data set, a charging and discharging benefit relationship of electric vehicles in each microgrid in the target multi-microgrid is obtained. The charging and discharging benefit relationship of electric vehicles is expressed as: in, Indicates the microgrids, Indicates the The charging and discharging benefits of electric vehicles in a microgrid, represents the total number of virtual energy storage of the first electric vehicles participating in charging scheduling, Indicates the Electric vehicles with virtual energy storage, express Time period The charging price of electric vehicles in a microgrid, express Time period In a microgrid The virtual energy storage charging capacity of electric vehicles.

[0032] A second data set of the hydrogen storage device of each microgrid is obtained based on the virtual energy storage relationship of the hydrogen storage device. The second data set includes at least: the daily hydrogen quality and the hydrogen production capacity of the alkaline electrolyzer. Based on the second data set, a hydrogen production efficiency relationship of each microgrid is obtained. The hydrogen production efficiency relationship is expressed as follows: in, Indicates the The hydrogen production benefits of a microgrid, represents the by-product profit coefficient of the hydrogen storage device, Indicates the hydrogen tank The hydrogen quality during the period, Indicates the daily mass of hydrogen, Represents the spot hydrogen price.

[0033] The power sold by each microgrid to other microgrids is obtained. Based on the power sold, the power selling benefit equation of each microgrid is obtained. The power selling benefit equation is expressed as: in, Indicates the The electricity sales benefit of a microgrid, express Time period The price at which a microgrid sells electricity to other microgrids, express Time period The power sold by a microgrid to other microgrids.

[0034] According to the relationship between electric vehicle charging and discharging benefits, hydrogen production benefits, and electricity sales benefits, a microgrid operation benefit model is constructed. The relationship between the microgrid operation benefit model is as follows: Furthermore, based on the electric vehicle virtual energy storage relationship, the wind turbine virtual energy storage relationship, the load control virtual energy storage relationship and the hydrogen storage device virtual energy storage relationship, an operating cost model for each microgrid in the target multi-microgrid is constructed. In a preferred embodiment of the present application, for each microgrid, considering the operating conditions of batteries, photovoltaics and other devices in the microgrid, a hydrogen storage device virtual energy storage scheduling cost relationship, a wind turbine virtual energy storage scheduling cost relationship, a controllable load virtual energy storage scheduling cost relationship, a hydrogen storage device virtual energy storage scheduling cost relationship, a microgrid network loss cost relationship, a battery cost relationship, a photovoltaic operation and maintenance cost relationship and a microgrid electricity purchase cost relationship are constructed. Specifically, based on the electric vehicle virtual energy storage relationship, a third data group of electric vehicle virtual energy storage participating in discharge scheduling of each microgrid is obtained. The third data group includes at least: the total number of second electric vehicle virtual energy storage participating in discharge scheduling and the amount of electric vehicle virtual energy storage discharged. Based on the third data group, the electric vehicle virtual energy storage scheduling cost relationship of each microgrid in the target multi-microgrid is obtained. The electric vehicle virtual energy storage scheduling cost relationship is expressed as follows: in, Indicates the The cost of dispatching electric vehicle virtual energy storage in a microgrid is represents the total number of virtual energy storage of the second electric vehicles participating in discharge scheduling, express Time period The electric vehicle virtual energy storage discharge compensation price of a microgrid, express Time period In a microgrid The amount of virtual energy storage discharged by an electric vehicle.

[0035] A fourth data set of wind turbine virtual energy storage wind regulation for each microgrid is obtained based on the wind turbine virtual energy storage relationship. The fourth data set includes at least: the total amount of wind turbine virtual energy storage, the output power of the wind turbine before wind regulation, and the output power of the wind turbine after wind regulation. Based on the fourth data set, a wind turbine virtual energy storage scheduling cost relationship is obtained for each microgrid. The wind turbine virtual energy storage scheduling cost relationship is expressed as follows: in, Indicates the The virtual energy storage dispatching cost of wind turbines in a microgrid is Indicates the total amount of virtual energy storage of wind turbines, Indicates the Virtual energy storage of wind turbines, Indicates the In the microgrid A fan adjustment factor, express Time period In the microgrid The power output of each wind turbine before virtual energy storage and wind adjustment, express Time period In the microgrid The power output of each wind turbine after virtual energy storage and wind adjustment, Indicates the In the microgrid The operation and maintenance cost of a wind turbine.

[0036] A fifth data set for the controllable load virtual energy storage participating in peak regulation of each microgrid is obtained based on the controllable load virtual energy storage relationship. The fifth data set includes at least: a power value of a curtailable load that can participate in scheduling, a power value of a transferable load that can participate in scheduling, and a power value of a shiftable load that can participate in scheduling. Based on the fifth data set, a controllable load virtual energy storage scheduling cost relationship is obtained for each microgrid. The controllable load virtual energy storage scheduling cost relationship is expressed as follows: in, Indicates the The virtual energy storage dispatching cost of the controllable load of a microgrid is express Time period The load power that can be reduced by the microgrid participating in the dispatch is Indicates the The compensation coefficient of virtual energy storage that can reduce load of a microgrid, express Time period The transferable load power of each microgrid participating in the dispatch, Indicates the The compensation coefficient of the transferable load virtual energy storage of a microgrid, express Time period The load power that can be shifted by the microgrid is: Indicates the The compensation coefficient of the virtual energy storage of the shiftable load of the microgrid.

[0037] A sixth data set of the hydrogen storage device virtual energy storage participating in peak load regulation of each microgrid is obtained based on the hydrogen storage device virtual energy storage relationship. The sixth data set includes at least: the power of the alkaline electrolyzer, the operating status of the alkaline electrolyzer, the power of the proton exchange membrane fuel cell, and the operating status of the proton exchange membrane fuel cell. Based on the sixth data set, a relationship formula for the virtual energy storage scheduling cost of the hydrogen storage device of each microgrid is obtained. The relationship formula for the virtual energy storage scheduling cost of the hydrogen storage device is expressed as follows: in, Indicates the The virtual energy storage dispatch cost of hydrogen storage devices in a microgrid, Indicates the total virtual energy storage of the hydrogen storage device, Indicates the A hydrogen storage device for virtual energy storage, Indicates the In a microgrid The water consumption of a hydrogen storage device for hydrogen production per day, Indicates the The unit water fee of a microgrid, express Time period In a microgrid The power of the alkaline electrolyzer of a hydrogen storage device, express Time period In a microgrid The operation of the alkaline electrolyzer of the hydrogen storage device, Indicates the In a microgrid The operation and maintenance costs of the alkaline electrolyzer of a hydrogen storage device, express Time period In a microgrid The power of the proton exchange membrane fuel cell with a hydrogen storage device, express Time period In a microgrid The operation of the proton exchange membrane fuel cell with a hydrogen storage device, Indicates the In a microgrid The operation and maintenance costs of a proton exchange membrane fuel cell with a hydrogen storage device.

[0038] Construct a microgrid network loss cost relationship, which is expressed as follows: in, Indicates the The network loss cost of a microgrid is express Time period The network loss of a microgrid, Indicates the The network loss cost coefficient of a microgrid.

[0039] Construct a battery cost relationship equation, which is expressed as follows: in, Indicates the The battery cost of a microgrid is Indicates the The total number of batteries in a microgrid, Indicates the A battery, express Time period In a microgrid The charge and discharge power of a battery, Indicates the In a microgrid The operation and maintenance cost of a battery, Indicates the In a microgrid The investment cost of a battery, Indicates the In a microgrid The number of charge and discharge times of a battery, Indicates the In a microgrid The depth of discharge of a battery, Indicates the In a microgrid The rated capacity of a battery.

[0040] Construct a photovoltaic operation and maintenance cost relationship formula, which is expressed as follows: in, Indicates the The photovoltaic operation and maintenance cost within a microgrid is Indicates the The total number of photovoltaic cells in a microgrid, Indicates the Photovoltaic, express Time period The first microgrid The output of photovoltaic Indicates the The first microgrid The operation and maintenance cost coefficient of each photovoltaic unit.

[0041] Construct the relationship between the cost of purchasing electricity for the microgrid. The relationship between the cost of purchasing electricity for the microgrid is as follows: in, Indicates the The electricity purchase cost of a microgrid is express Time period The amount of electricity purchased by a microgrid, express Time period The electricity purchase price of a microgrid.

[0042] Furthermore, based on the virtual energy storage dispatching cost relationship of electric vehicles, the virtual energy storage dispatching cost relationship of wind turbines, the virtual energy storage dispatching cost relationship of controllable loads, the virtual energy storage dispatching cost relationship of hydrogen storage devices, the microgrid network loss cost relationship, the battery cost relationship, the photovoltaic operation and maintenance cost relationship, and the microgrid electricity purchase cost relationship, a microgrid operation cost model for each microgrid is constructed. The relationship of the microgrid operation cost model is expressed as follows: S2. Constructing a microgrid optimization model for each microgrid based on the output results corresponding to the operation benefit model and the operation cost model, and optimizing the target multi-microgrid based on the microgrid optimization model of each microgrid to obtain a multi-microgrid optimization model. In a preferred embodiment of the present application, a microgrid optimization model is constructed for each microgrid. The microgrid optimization model is determined by the relational expression of the microgrid operation benefit model and the relational expression of the microgrid operation cost model of the microgrid. The relational expression of the microgrid optimization model is expressed as: in, Indicates the Optimization benefits of a microgrid.

[0043] Furthermore, taking the minimum difference between the multi-microgrid operating benefits and operating costs as the optimization goal, an optimization objective function is constructed, and a multi-microgrid optimization model is constructed based on the optimization objective function. The objective function expression of the multi-microgrid optimization model is: in, represents the optimization objective function of the multi-microgrid optimization model, Represents the collection of microgrids contained in the multi-microgrid.

[0044] The various constraints of the multi-microgrid optimization model are expressed as: in, and They represent the upper and lower limits of the virtual energy storage charge state of electric vehicles, and They represent the upper and lower limits of the aging degree of electric vehicle batteries, and They represent the upper and lower limits of the wind turbine virtual energy storage charge state, Indicates the upper limit of load reduction. and Represent the upper and lower limits of the transferable load, and Respectively represent the upper and lower limits of the translatable load, and They represent the upper and lower limits of the virtual energy storage charge state of the hydrogen storage device, express The state of charge of the battery during the period, and Respectively represent the upper and lower limits of the battery's state of charge, Indicates the line power, Indicates the maximum transmission power of the line, and Respectively Time period The battery charging power and discharging power in each microgrid, express Time period The power of the inherent load of a microgrid.

[0045] The multi-microgrid optimization model constructed in this application virtualizes electric vehicles, wind turbines, controllable loads, and hydrogen storage devices as energy storage. Using virtual capacitance values ​​as standardized energy regulation parameters, this streamlines the operating parameters of the multi-microgrid's flexible and adjustable resources, thereby improving the perception of the multi-microgrid's energy regulation potential. Using the virtual state of charge as the status criterion for virtual energy storage provides a unified quantitative benchmark for evaluating the frequency and peak regulation capabilities of multi-microgrids.

[0046] S3. Based on the microgrid optimization model and the multi-microgrid optimization model, an initial Markov game model is constructed, and local model training is performed on the initial Markov game model to obtain the noise gradient of each agent. The initial Markov game model defines the agent based on each microgrid, and the microgrid optimization model is converted into the local model, and the multi-microgrid optimization model is converted into a global model. In a preferred embodiment of the present application, a Markov game algorithm is used to perform multivariate scheduling on the multi-microgrid, defining each microgrid as an agent, constructing a local model based on the microgrid optimization model of each microgrid, and constructing a global model based on the multi-microgrid optimization model. A Markov game algorithm is a mathematical model used to describe the interactive decision-making of multiple agents in a dynamic environment. The basic elements of a Markov game algorithm include a state space, an action space, a transition probability, and a reward function. The state space contains all possible states of the game objects in the game. In a preferred embodiment of the present application, the overall observation information of each microgrid is the same, while the local observation information is different. A state space is constructed from the local and overall microgrid observation information. The global state space focuses on the unit operation and maintenance cost and time of each device, including the spot hydrogen price, unit water fee, compensation coefficient of virtual energy storage for load reduction, compensation coefficient of virtual energy storage for load transfer, compensation coefficient of virtual energy storage for load translation, electricity purchase price and time period. The local state space focuses on the electricity sales price and electric vehicle charging and discharging price, photovoltaic and wind turbine output, virtual energy storage state of charge and inherent load of each microgrid, including the electricity sales price to other microgrids, electric vehicle charging price, electric vehicle virtual energy storage discharge compensation price, photovoltaic output, battery state of charge, electric vehicle virtual energy storage state of charge, electric vehicle aging degree, wind turbine virtual energy storage state of charge, and hydrogen storage device virtual energy storage state of charge.

[0047] The state space is represented as: in, Indicates the The local state space of a microgrid, Indicates the The global state space of a microgrid, Indicates the The state space of a microgrid.

[0048] Furthermore, the action space of each microgrid is constructed. Based on the optimizable variables of various devices in each microgrid, namely the charge and discharge power of various virtual energy storages, the charge and discharge power of batteries, and the power purchased and sold by the microgrid, the action space of multiple microgrids is established. The action space is expressed as follows: Furthermore, a reward function for the multi-microgrid is constructed. In a preferred embodiment of the present application, the reward function includes a reward for achieving the optimization goal and a penalty reward for exceeding the constraint limit. Taking into full consideration the optimization objective function and constraint conditions of the multi-microgrid, the expression of the reward function is: in, represents the reward function, is the weight coefficient of the optimization objective of the multi-microgrid optimization model, is the weight coefficient of the constraint violation penalty, Represents the penalty cost of exceeding the limit for various constraints in the multi-microgrid optimization model.

[0049] Based on the state space, action space, reward function, microgrid optimization model, and multi-microgrid optimization model, an initial Markov game model is constructed. This initial Markov game model includes a local model based on the microgrid optimization model and a global model based on the multi-microgrid optimization model. The state of the initial Markov game model at the next moment depends only on the current state and the action taken, and is independent of past historical states. This enables the model to effectively handle decision-making problems in dynamic environments and simplifies the complexity of the problem. Considering the interactions between multiple agents, each agent's decision affects not only its own reward but also the rewards of other agents and the state transition of the entire multi-microgrid. The Markov game model effectively describes the dynamic changes of the multi-microgrid over time and how the actions of a particular microgrid affect the state transition of the multi-microgrid. Different scheduling strategies are evaluated and compared to obtain the optimal scheduling strategy for each microgrid under different states, achieving rational resource allocation and improving the resource utilization efficiency of the multi-microgrid.

[0050] Furthermore, a deep deterministic policy gradient reinforcement learning method is used for local model training. Specifically, the parameters of each agent's actor network and critic network are initialized. The actor network is responsible for generating action policies, while the critic network evaluates the value of the action policies generated by the actor network, obtaining reward feedback, which is fed back to the actor network for policy optimization. Each agent obtains its own state space from the environment as input. The actor network performs a set of actions and sends them to the critic network, receiving the corresponding reward feedback and post-action state. The state space, actions, reward feedback, and post-action state are combined as experience samples and stored in the experience buffer pool. Each agent then draws small batches of experience samples from the experience buffer pool for training and model updates, and each agent calculates the latest model gradient.

[0051] In the preferred embodiment of this application, in order to protect the privacy of the local model training results, according to the Laplace mechanism in differential privacy, Laplace noise is introduced into the latest gradient to obtain the noise gradient of each agent. The noise gradient is expressed as follows: in, Indicates the The noise gradient of each agent, After the noise is introduced Actor network parameters, After the noise is introduced Critic network parameters of each agent, Indicates the The initial Actor network parameters of each agent, Indicates the The initial critic network parameters of each agent, Indicates the Laplace noise of the agents.

[0052] S4. Using a federated learning algorithm to update the noise gradient of each agent, and iteratively training the local model based on the obtained global model noise gradient until the local model meets preset conditions and obtains the target Markov game model. In a preferred embodiment of the present application, a federated learning algorithm is used for global model aggregation. According to the federated learning training framework, the noise gradient of each agent is uploaded to the global server to update the global model parameters and obtain the global model noise gradient. The global model noise gradient is globally broadcasted to update the noise gradient of each agent's local model, and iterative training is performed based on the updated noise gradient until the local model loss curve and / or the local model gain curve converge, thereby obtaining the target Markov game model.

[0053] After obtaining the target Markov game model, the first real-time operating data of the virtual energy storage and the second real-time operating data of the target multi-microgrid are input into the target Markov game model to obtain an optimized scheduling strategy, and the target multi-microgrid is scheduled according to the optimized scheduling strategy. The first real-time operating data includes at least the parameters involved in constructing the virtual energy storage equations for electric vehicles, wind turbines, controllable loads, and hydrogen storage devices. The second real-time operating data includes at least the parameters involved in constructing the microgrid network loss cost equations, battery cost equations, photovoltaic operation and maintenance cost equations, and microgrid electricity purchase cost equations. This fully leverages the synergistic potential of multi-microgrid sources, grids, loads, and storage. The distributed renewable energy configuration on the power source side and the various distributed controllable loads on the user side of each microgrid are perceived, enabling unified multi-dimensional energy scheduling across the multi-microgrids. This meets the requirements of multi-microgrid coordinated control, significantly improves the benefits of coordinated optimization, achieves rational resource allocation, improves the utilization efficiency of multi-microgrid resources, and ensures data security for each microgrid.

[0054] In a preferred embodiment of the present invention, the virtual energy storage type of the target multi-microgrid is determined, and the first real-time operation data of each type of virtual energy storage and the second real-time operation data of the target multi-microgrid are input into the target Markov game model, and the target multi-microgrid is scheduled according to the obtained scheduling strategy. The construction and training process of the target Markov game model includes: constructing an operation model of each microgrid in the target multi-microgrid according to the analysis results of the operation data of each type of virtual energy storage, and the operation model includes an operation benefit model and an operation cost model; constructing a microgrid optimization model of each microgrid according to the output results corresponding to the operation benefit model and the operation cost model, and constructing a microgrid optimization model of each microgrid according to the microgrid optimization results of each microgrid. The model optimizes the target multi-microgrid to obtain a multi-microgrid optimization model. Based on the microgrid optimization model and the multi-microgrid optimization model, an initial Markov game model is constructed and locally trained to obtain the noise gradient of each agent. The initial Markov game model defines agents for each microgrid, converts the microgrid optimization model into a local model, and converts the multi-microgrid optimization model into a global model. A federated learning algorithm is used to update the noise gradient of each agent, and the local model is iteratively trained based on the obtained global model noise gradient until the local model meets preset conditions, thereby obtaining the target Markov game model. The multi-microgrid multivariate scheduling method based on virtual energy storage disclosed in this application virtualizes electric vehicles, wind turbines, controllable loads, and hydrogen storage devices into energy storage forms, using virtual capacitance values ​​as standardized energy control parameters to streamline the operating parameters of the multi-microgrid's flexible adjustable resources, thereby improving the perception of the multi-microgrid's energy regulation potential. Using the virtual state of charge as the state criterion for virtual energy storage provides a unified quantitative benchmark for evaluating the frequency and peak regulation capabilities of multi-microgrids. The Markov game algorithm fully taps the collaborative potential of multiple microgrid sources, grids, loads and storage, perceives the distributed renewable energy configuration on the power supply side of each microgrid and various distributed controllable loads on the user side, realizes unified multi-dimensional energy scheduling of multiple microgrids, meets the needs of coordinated regulation of multiple microgrids, significantly improves the collaborative optimization benefits, realizes the rational allocation of resources, improves the utilization efficiency of multiple microgrid resources, and ensures the data security of each microgrid.

[0055] Accordingly, if Figure 2 The figure shows a structural diagram of a multi-microgrid multi-dispatching system based on virtual energy storage. Based on a multi-microgrid multi-dispatching method based on virtual energy storage, an embodiment of the present invention further provides a multi-microgrid multi-dispatching system based on virtual energy storage, which implements the multi-microgrid multi-dispatching method based on virtual energy storage disclosed in an embodiment of the present invention. The system includes: a multi-dispatching module; The multivariate scheduling module is used to determine the virtual energy storage type of the target multi-microgrid, input the first real-time operation data of each type of virtual energy storage and the second real-time operation data of the target multi-microgrid into the target Markov game model, and schedule the target multi-microgrid according to the obtained scheduling strategy; The multivariate dispatch module includes: a microgrid operation model construction unit 1, a multi-microgrid optimization model construction unit 2, an initial Markov game model construction unit 3 and a model iterative training unit 4; The microgrid operation model construction unit 1 is used to construct an operation model of each microgrid in the target multi-microgrid according to the operation data analysis results of each type of virtual energy storage, and the operation model includes an operation benefit model and an operation cost model; The multi-microgrid optimization model construction unit 2 is used to construct a microgrid optimization model for each of the microgrids according to the output results corresponding to the operation benefit model and the operation cost model, and optimize the target multi-microgrid according to the microgrid optimization model of each of the microgrids to obtain a multi-microgrid optimization model; The initial Markov game model construction unit 3 is used to construct an initial Markov game model based on the microgrid optimization model and the multi-microgrid optimization model, and perform local model training on the initial Markov game model to obtain the noise gradient of each intelligent agent. The initial Markov game model defines the intelligent agent in each microgrid, and converts the microgrid optimization model into the local model and converts the multi-microgrid optimization model into a global model. The model iterative training unit 4 is used to update the noise gradient of each intelligent agent using a federated learning algorithm, and iteratively train the local model according to the obtained global model noise gradient until the local model meets the preset conditions to obtain the target Markov game model.

[0056] For the specific definition of a multi-microgrid multi-dispatching system based on virtual energy storage, please refer to the above-mentioned definition of a multi-microgrid multi-dispatching method based on virtual energy storage, which will not be repeated here. A person of ordinary skill in the art will appreciate that the various modules and steps described in conjunction with the embodiments disclosed in the present invention can be implemented in hardware, software, or a combination of both. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present invention.

[0057] like Figure 3 As shown, an embodiment of the present invention provides a computer device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, the steps in the multi-microgrid multi-element scheduling embodiment based on virtual energy storage are implemented, for example Figure 1 Steps S1 to S4 described in .

[0058] Those skilled in the art will understand that the schematic Figure 3 These are merely examples of computer devices and do not constitute limitations on the computer device. The computer device may include more or fewer components than shown in the figure, or a combination of certain components, or different components. For example, the computer device may also include input and output devices, network access devices, buses, etc.

[0059] The processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the computer device, connecting various parts of the entire computer device using various interfaces and lines.

[0060] The memory can be used to store the computer programs and / or modules. The processor implements the various functions of the computer device by running or executing the computer programs and / or modules stored in the memory and accessing the data stored in the memory. The memory may mainly include a program storage area and a data storage area. The program storage area may store an operating system and at least one application required for a function (such as a sound playback function, an image playback function, etc.); the data storage area may store data generated based on the use of the mobile phone (such as audio data, a phone book, etc.). In addition, the memory may include high-speed random access memory and non-volatile memory, such as a hard disk, internal memory, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, at least one disk storage device, a flash memory device, or other volatile solid-state storage device.

[0061] If the module integrated into the computer device is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present invention can also implement all or part of the process steps in the above-mentioned method embodiments by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When executed by a processor, the computer program can implement the steps of each of the above-mentioned method embodiments. The computer program includes computer program code, which can be in source code form, object code form, executable file, or some intermediate form. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard drive, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium.

[0062] Those skilled in the art will appreciate that all or part of the processes in the above-described method embodiments can be implemented by instructing the relevant hardware through a computer program. The program can be stored in a computer-readable storage medium, and when executed, the program can include the processes in the above-described method embodiments. The storage medium can be a magnetic disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).

[0063] Accordingly, an embodiment of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium includes a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to perform the steps in the multi-microgrid multi-element scheduling based on virtual energy storage as described in the above embodiment, for example Figure 1 Steps S1 to S4 described in .

[0064] In summary, the embodiment of the present application provides a multi-microgrid multi-element scheduling method, system, device and medium based on virtual energy storage, which solves the technical problems of how to achieve multi-element optimization scheduling of multi-microgrids and improve the data security of multi-microgrids. The method includes: determining the virtual energy storage type of the target multi-microgrid, and inputting the first real-time operation data of each type of virtual energy storage and the second real-time operation data of the target multi-microgrid into the target Markov game model, and scheduling the target multi-microgrid according to the obtained scheduling strategy. The construction and training process of the target Markov game model includes: constructing an operation model of each microgrid in the target multi-microgrid according to the analysis results of the operation data of each type of virtual energy storage, and the operation model includes an operation benefit model and an operation cost model; The output results are used to construct a microgrid optimization model for each microgrid, and the target multi-microgrid is optimized based on the microgrid optimization model of each microgrid to obtain a multi-microgrid optimization model. An initial Markov game model is constructed based on the microgrid optimization model and the multi-microgrid optimization model, and the initial Markov game model is trained locally to obtain the noise gradient of each agent. The initial Markov game model defines the agent in each microgrid, and the microgrid optimization model is converted into a local model, and the multi-microgrid optimization model is converted into a global model. The noise gradient of each agent is updated using a federated learning algorithm, and the local model is iteratively trained based on the obtained global model noise gradient until the local model meets the preset conditions, thereby obtaining the target Markov game model. The multi-microgrid multi-element scheduling method based on virtual energy storage disclosed in this application virtualizes electric vehicles, wind turbines, controllable loads, and hydrogen storage devices into energy storage forms, uses virtual capacitance values ​​as standardized energy control parameters, simplifies the operating parameters of the flexible adjustable resources of the multi-microgrid, and thus improves the perception level of the energy regulation potential of the multi-microgrid. Using the virtual state of charge as a criterion for virtual energy storage provides a unified quantitative benchmark for evaluating the frequency and peak regulation capabilities of multiple microgrids. The Markov game algorithm fully leverages the synergistic potential of multiple microgrids' sources, grids, loads, and storage. It perceives the distributed renewable energy configuration on the power supply side of each microgrid and the various distributed controllable loads on the user side, enabling unified, multi-dimensional energy scheduling across multiple microgrids. This meets the needs of multi-microgrid coordinated control, significantly improves collaborative optimization benefits, achieves rational resource allocation, improves multi-microgrid resource utilization efficiency, and ensures data security within each microgrid.

[0065] Each embodiment in this specification is described in a progressive manner, and the same or similar parts of each embodiment can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment. It should be noted that the various technical features of the above embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the various technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0066] The above-described embodiments merely represent several preferred implementations of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the patent application. It should be noted that a person skilled in the art could make several improvements and substitutions without departing from the technical principles of the present application, and such improvements and substitutions should also be considered within the scope of protection of the present application. Therefore, the scope of protection of the present patent application shall be based on the scope of protection of the claims.

Claims

1. A multi-microgrid multi-element dispatching method based on virtual energy storage, characterized in that: The method comprises: Determine the virtual energy storage type of the target multi-microgrid, input the first real-time operation data of each type of virtual energy storage and the second real-time operation data of the target multi-microgrid into the target Markov game model, and schedule the target multi-microgrid according to the obtained scheduling strategy. The construction and training process of the target Markov game model includes: Constructing an operation model of each microgrid in the target multi-microgrid according to the analysis results of the operation data of each type of virtual energy storage, wherein the operation model includes an operation benefit model and an operation cost model; Constructing a microgrid optimization model for each microgrid according to the output results corresponding to the operation benefit model and the operation cost model, and optimizing the target multi-microgrid according to the microgrid optimization model of each microgrid to obtain a multi-microgrid optimization model; Constructing an initial Markov game model based on the microgrid optimization model and the multi-microgrid optimization model, and performing local model training on the initial Markov game model to obtain a noise gradient for each intelligent agent, wherein the initial Markov game model defines the intelligent agent based on each microgrid, and converting the microgrid optimization model into the local model and converting the multi-microgrid optimization model into a global model; A federated learning algorithm is used to update the noise gradient of each agent, and the local model is iteratively trained according to the obtained global model noise gradient until the local model meets the preset conditions, thereby obtaining the target Markov game model.

2. The multi-microgrid multi-element dispatching method based on virtual energy storage according to claim 1, characterized in that: The operation model of each microgrid in the target multi-microgrid is constructed according to the analysis results of the operation data of each type of virtual energy storage, including: Based on the switching characteristics of electric vehicles under the guidance of time-of-use electricity prices, electric vehicle operation data is analyzed to construct an electric vehicle virtual energy storage relationship. The electric vehicle virtual energy storage relationship includes: an electric vehicle virtual capacitance value relationship, an electric vehicle virtual state of charge relationship, and an electric vehicle battery aging relationship. The process of controlling the rotor kinetic energy by adjusting the motor state of the wind turbine is equivalent to the charging and discharging process of the wind turbine, and the wind turbine operating data is analyzed to construct a wind turbine virtual energy storage relationship, which includes: a wind turbine virtual capacitance value relationship and a wind turbine virtual state of charge relationship; The response process of the curtailable load, transferable load, and shiftable load of the controllable load is equated to the charging and discharging process of the controllable load, and the controllable load operation data is analyzed to construct a virtual energy storage relationship of the controllable load, wherein the virtual energy storage relationship of the controllable load includes: a virtual power value relationship of the curtailable load, a virtual power value relationship of the transferable load, and a virtual power value relationship of the shiftable load; According to the energy conversion process of the hydrogen storage device, the operating data of the hydrogen storage device is analyzed to construct a virtual energy storage relationship of the hydrogen storage device, which includes: a relationship between the amount of hydrogen produced by the alkaline electrolyzer, a relationship between the amount of hydrogen burned by the proton exchange membrane fuel cell, a relationship between the virtual state of charge of the hydrogen storage tank, and a relationship between the virtual capacitance value of the hydrogen storage device; Constructing an operation benefit model for each microgrid in the target multi-microgrid according to the electric vehicle virtual energy storage relationship and the hydrogen storage device virtual energy storage relationship; An operation cost model of each microgrid in the target multi-microgrid is constructed based on the electric vehicle virtual energy storage relationship formula, the wind turbine virtual energy storage relationship formula, the controllable load virtual energy storage relationship formula and the hydrogen storage device virtual energy storage relationship formula.

3. The multi-microgrid multi-element dispatching method based on virtual energy storage according to claim 2, characterized in that: The constructing of an operation benefit model for each microgrid in the target multi-microgrid according to the electric vehicle virtual energy storage relationship and the hydrogen storage device virtual energy storage relationship includes: Obtaining a first data set of electric vehicle virtual energy storage participating in charging scheduling for each microgrid within the target multi-microgrid according to the electric vehicle virtual energy storage relationship formula, and obtaining a charging and discharging benefit relationship formula for each microgrid within the target multi-microgrid based on the first data set, wherein the first data set includes at least: a total number of first electric vehicle virtual energy storage participating in charging scheduling and a charging amount of the electric vehicle virtual energy storage; Obtaining a second data set of virtual energy storage of the hydrogen storage device of each microgrid according to the virtual energy storage relationship of the hydrogen storage device, and obtaining a hydrogen production benefit relationship of each microgrid based on the second data set, wherein the second data set includes at least: daily hydrogen quality and hydrogen production of the alkaline electrolyzer; Obtaining the power sold by each microgrid to other microgrids, and obtaining a power sales benefit relationship formula for each microgrid based on the power sold; An operation benefit model of each microgrid in the target multi-microgrid is constructed based on the electric vehicle charging and discharging benefit relationship, the hydrogen production benefit relationship, and the electricity sales benefit relationship.

4. The multi-microgrid multi-element dispatching method based on virtual energy storage according to claim 2, characterized in that: The operation cost model of each microgrid in the target multi-microgrid is constructed according to the electric vehicle virtual energy storage relationship, the wind turbine virtual energy storage relationship, the controllable load virtual energy storage relationship, and the hydrogen storage device virtual energy storage relationship, including: Obtaining a third data set of electric vehicle virtual energy storage participating in discharge scheduling for each microgrid within the target multi-microgrid based on the electric vehicle virtual energy storage relationship formula, and obtaining a scheduling cost relationship formula for the electric vehicle virtual energy storage for each microgrid within the target multi-microgrid based on the third data set, wherein the third data set includes at least: a total number of second electric vehicle virtual energy storage participating in discharge scheduling and a discharge amount of the electric vehicle virtual energy storage; Obtaining a fourth data group of wind turbine virtual energy storage wind regulation of each microgrid according to the wind turbine virtual energy storage relationship formula, and obtaining a wind turbine virtual energy storage scheduling cost relationship formula for each microgrid based on the fourth data group, wherein the fourth data group includes at least: the total number of wind turbine virtual energy storage, the output power of the wind turbine virtual energy storage before wind regulation, and the output power of the wind turbine virtual energy storage after wind regulation; Obtaining a fifth data group for the controllable load virtual energy storage of each microgrid participating in peak regulation based on the controllable load virtual energy storage relationship formula, and obtaining a controllable load virtual energy storage scheduling cost relationship formula for each microgrid based on the fifth data group, wherein the fifth data group includes at least: a power value of a curtailable load that can participate in scheduling, a power value of a transferable load that can participate in scheduling, and a power value of a shiftable load that can participate in scheduling; Obtaining a sixth data set of the virtual energy storage participation of the hydrogen storage device of each microgrid in peak load regulation according to the hydrogen storage device virtual energy storage relationship formula, and obtaining a virtual energy storage scheduling cost relationship formula of the hydrogen storage device of each microgrid based on the sixth data set, wherein the sixth data includes at least: the power of the alkaline electrolyzer, the operating status of the alkaline electrolyzer, the power of the proton exchange membrane fuel cell, and the operating status of the proton exchange membrane fuel cell; Obtaining other operating cost relationship formulas for each of the microgrids, wherein the other operating costs include at least: a microgrid network loss cost relationship formula, a battery cost relationship formula, a photovoltaic operation and maintenance cost relationship formula, and a microgrid electricity purchase cost relationship formula; According to the electric vehicle virtual energy storage scheduling cost relationship formula, the wind turbine virtual energy storage scheduling cost relationship formula, the controllable load virtual energy storage scheduling cost relationship formula, the hydrogen storage device virtual energy storage scheduling cost relationship formula and the other operating cost relationship formulas, an operating cost model of each microgrid in the target multi-microgrid is constructed.

5. The multi-microgrid multi-element dispatching method based on virtual energy storage according to claim 1, characterized in that: The method of constructing an initial Markov game model based on the microgrid optimization model and the multi-microgrid optimization model, and performing local model training on the initial Markov game model to obtain the noise gradient of each intelligent agent includes: Defining each microgrid in the target multi-microgrid as an intelligent agent, obtaining a state space for observing each microgrid, wherein the state space includes an overall state space for observing the microgrid and a local state space for observing the microgrid; Obtaining an action space of each of the intelligent agents, the action space including at least: a first charge and discharge power of each type of virtual energy storage, a second charge and discharge power of the battery, and a purchase and sale power of the microgrid; Constructing a reward function for the target multi-microgrid according to the multi-microgrid optimization model, wherein the reward function includes: an optimization goal achievement reward and a constraint violation penalty reward; Converting the microgrid optimization model into a local model and converting the multi-microgrid optimization model into a global model; constructing an initial Markov game model according to the state space, the action space, the reward function, the local model, and the global model; A deep deterministic policy gradient reinforcement learning method is used to perform local model training on the initial Markov game model to obtain the latest gradient of each local model, and Laplace noise is introduced into the latest gradient to obtain the noise gradient of each intelligent agent.

6. The multi-microgrid multi-element dispatching method based on virtual energy storage according to claim 5, characterized in that: The reward function is expressed as: in, represents the reward function, represents the weight coefficient of the optimization objective of the multi-microgrid optimization model, represents the optimization objective function of the multi-microgrid optimization model, represents the weight coefficient of the constraint violation penalty, Represents the penalty cost of exceeding the limit for various constraints in the multi-microgrid optimization model.

7. The multi-microgrid multi-element dispatching method based on virtual energy storage according to claim 1, characterized in that: The federated learning algorithm is used to update the noise gradient of each agent, and the local model is iteratively trained according to the obtained global model noise gradient until the local model meets the preset conditions to obtain the target Markov game model, including: Uploading the noise gradient of each agent to a global server to update global model parameters to obtain a global model noise gradient; The global model noise gradient is globally broadcasted to update the noise gradient of each of the intelligent agents, and iterative training is performed based on the updated noise gradient until the local model meets preset conditions to obtain the target Markov game model. The preset conditions include: the local model loss curve converges and / or the local model gain curve converges.

8. A multi-microgrid multi-element dispatching system based on virtual energy storage, used to implement the multi-microgrid multi-element dispatching method based on virtual energy storage according to any one of claims 1 to 7, characterized in that: The system includes: a multivariate scheduling module; The multivariate scheduling module is used to determine the virtual energy storage type of the target multi-microgrid, input the first real-time operation data of each type of virtual energy storage and the second real-time operation data of the target multi-microgrid into the target Markov game model, and schedule the target multi-microgrid according to the obtained scheduling strategy; The multivariate dispatch module includes: a microgrid operation model construction unit, a multi-microgrid optimization model construction unit, an initial Markov game model construction unit and a model iteration training unit; The microgrid operation model construction unit is used to construct an operation model of each microgrid in the target multi-microgrid according to the operation data analysis results of each type of virtual energy storage, wherein the operation model includes an operation benefit model and an operation cost model; The multi-microgrid optimization model construction unit is configured to construct a microgrid optimization model for each microgrid according to the output results of the operation benefit model and the operation cost model, and optimize the target multi-microgrid according to the microgrid optimization model of each microgrid to obtain a multi-microgrid optimization model; The initial Markov game model construction unit is configured to construct an initial Markov game model based on the microgrid optimization model and the multi-microgrid optimization model, and perform local model training on the initial Markov game model to obtain a noise gradient of each intelligent agent. The initial Markov game model defines the intelligent agent based on each microgrid, and converts the microgrid optimization model into the local model and converts the multi-microgrid optimization model into a global model. The model iterative training unit is used to update the noise gradient of each intelligent agent using a federated learning algorithm, and iteratively train the local model according to the obtained global model noise gradient until the local model meets the preset conditions, thereby obtaining the target Markov game model.

9. A computer device, characterized in that: The computer device includes a memory, a processor and a transceiver, which are connected via a bus; the memory is used to store a set of computer program instructions and data, and transmit the stored data to the processor, and the processor executes the computer program instructions stored in the memory to perform the multi-microgrid multi-element scheduling method based on virtual energy storage as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed, the multi-microgrid multi-element scheduling method based on virtual energy storage according to any one of claims 1 to 7 is implemented.

Images