Virtual power plant configuration optimization method and system based on multi-agent cooperative game

Abstract

The invention relates to the technical field of virtual power plant operation, in particular to a virtual power plant configuration optimization method and system for a multi-agent cooperative game. According to the method, the cooperation value of each main body is evaluated by analyzing the difference between the individual operation income and the cooperation operation income of each main body in the virtual power plant; determining the dynamic responsivity according to the current power of the main body through the power fluctuation deviation of the main body in the historical operation, and determining the contribution ability of the main body in combination with the loss risk possibly borne by each main body in the operation of the virtual power plant; and optimizing configuration operation based on matching of the dynamic contribution capability and the cooperation value. According to the method, a contribution perception optimization overall configuration scheme is introduced through the matching relationship between the contribution and the obtained income of each main body in the collaborative participation process, so that the operation efficiency of the virtual power plant, the incentive effectiveness of income distribution and the stability of the multi-main body alliance are improved, and the structural disjunction problem between configuration optimization and income distribution is reduced.

Description

Technical Field

[0001] This invention relates to the field of virtual power plant operation technology, specifically to a method and system for optimizing the configuration of virtual power plants through multi-party collaborative game theory. Background Technology

[0002] With the rapid development of new energy and distributed energy, traditional centralized power systems are gradually transforming into distributed and intelligent systems. Virtual power plants (VPPs), as a new energy management approach, use information communication and control technologies to virtually aggregate various energy resources such as photovoltaics, wind power, energy storage, and electric vehicles located in different locations for unified scheduling and control, achieving the function of "looking like a large power plant." Multiple stakeholders exist within a virtual power plant, such as energy producers, energy storage operators, and demand response users, each with different objective functions and operational preferences, leading to resource competition, operational conflicts, or contradictions in revenue distribution. Game theory can be used to simulate and solve strategic interaction problems among multiple stakeholders. Within the framework of cooperative game theory, through alliance modeling, marginal contribution calculation, and revenue distribution strategies, a cooperative mechanism can be established among stakeholders to collaboratively optimize energy resource allocation, improve overall system efficiency, and ensure individual benefits.

[0003] In the actual operation of virtual power plants, participating entities exhibit significant dynamic changes and asymmetries in terms of resource capabilities, control response behavior, and cost structure. Different entities differ in their adjustable resource scale, load response frequency, equipment availability, and ability to withstand market fluctuations. However, current mainstream configuration optimization methods generally prioritize maximizing total system revenue, failing to effectively incorporate individual differences in dispatch frequency, risk-bearing capacity, and actual output capacity. This results in some entities bearing excessive burdens while receiving insufficient returns, leading to a mismatch between revenue distribution and actual contribution, thus undermining incentive mechanisms and alliance stability. Summary of the Invention

[0004] To address the technical problems in existing technologies where traditional configuration optimization methods focus on maximizing system benefits while neglecting differences in individual participation and contribution, leading to unfair benefit distribution and incentive failure, the present invention aims to provide a virtual power plant configuration optimization method and system based on multi-agent collaborative game theory. The specific technical solution adopted is as follows: This invention provides a method for optimizing the configuration of a virtual power plant through multi-agent collaborative game theory, the method comprising: Obtain the collaborative benefits and actual power of each entity during each resource scheduling cycle in the collaborative runtime of the virtual power plant; Based on the independent operating benefits and collaborative benefits of a single entity in the current resource scheduling cycle, analyze the current collaborative value of the entity; During the collaborative operation of a single entity, the fluctuation deviation coefficient of the entity is determined by the degree of change between the actual power fluctuation and the expected power supply deviation in consecutive adjacent resource scheduling cycles. Based on the actual power fluctuation and expected power in the current resource scheduling cycle of a single entity, the current dynamic responsiveness of the entity is obtained by combining the fluctuation deviation coefficient. Based on the equipment loss of a single entity's equipment in the current resource scheduling cycle and its relationship with other entities, analyze the current risk compensation coefficient of the single entity; By analyzing the current dynamic responsiveness and risk compensation coefficient of the subject, we can determine the current dynamic contribution capability of the subject; based on the matching relationship between the current dynamic contribution capability and the collaborative value of a single subject, we can obtain the contribution-return matching degree of the subject. Configuration schemes are obtained through distributed optimization based on the matching degree of contribution and benefit.

[0005] Furthermore, the method for obtaining the collaborative value includes: Obtain the independent revenue of each entity operating independently within the current resource scheduling cycle; The difference between the collaborative gains and independent gains of each entity within the current resource scheduling cycle is normalized to obtain the current collaborative value of each entity.

[0006] Furthermore, the method for obtaining the fluctuation deviation coefficient includes: For any entity, the resource volatility of the entity in each resource scheduling cycle is obtained based on the degree of fluctuation and disorder of the actual power of the entity in each resource scheduling cycle and the deviation of the actual power fluctuation from the resource scheduling cycle in the cooperative runtime. Calculate the difference between the actual power and the expected power of the entity at each moment in each resource scheduling cycle, and normalize the mean of all differences to obtain the resource supply variability of the entity in each resource scheduling cycle. During the collaborative operation period of this entity, the ratio of the difference in resource volatility to the difference in resource supply between any two adjacent resource scheduling cycles is taken as the adjacent change ratio between any two adjacent resource scheduling cycles; the average of the adjacent change ratios between all two adjacent resource scheduling cycles during the collaborative operation period is taken as the volatility deviation coefficient of this entity.

[0007] Furthermore, the method for obtaining resource volatility includes: The variance of the actual power of the subject in each resource scheduling cycle is taken as the degree of disorder variation in each resource scheduling cycle; the mean of the degree of disorder variation of the subject in all resource scheduling cycles during the cooperative runtime is taken as the mean of the subject's variation. Calculate the difference between the degree of disorder and the mean of change for each resource scheduling cycle, and use the ratio of the difference to the mean of change as the degree of fluctuation deviation for each resource scheduling cycle. The resource volatility of the entity is obtained by normalizing the product of the fluctuation deviation and the disorder variation in each resource scheduling cycle.

[0008] Furthermore, the method for obtaining the dynamic response degree includes: For any subject, the actual power variance of the subject within the current resource scheduling cycle is taken as the current power volatility of the subject; the product of the subject's current volatility and volatility deviation coefficient is taken as the subject's current response deviation. The difference between the expected power of the subject in the current resource scheduling cycle and the current response deviation is normalized to obtain the current dynamic response of the subject.

[0009] Furthermore, the method for obtaining the risk compensation coefficient includes: Obtain the equipment loss of each entity under the expected power in the current resource scheduling cycle; The difference between the equipment loss of each entity and the average loss of all entities' equipment is normalized to obtain the risk compensation coefficient for each entity.

[0010] Furthermore, the method for acquiring the dynamic contribution capability includes: For any given entity, the product of its risk compensation coefficient and dynamic responsiveness is taken as its risk adjustment degree; the sum of its risk adjustment degree and dynamic responsiveness is taken as its dynamic contribution capability.

[0011] Furthermore, the method for obtaining the contribution-benefit matching degree includes: By negatively mapping and normalizing the difference between the current dynamic contribution capacity and collaborative value of a single entity, the contribution-benefit matching degree of the entity is obtained.

[0012] Furthermore, the configuration scheme obtained through distributed optimization based on contribution-benefit matching degree includes: The optimal configuration scheme is obtained by using the matching degree of each subject's contribution and benefit as a weight and collaboratively solving the problem through the ADMM algorithm.

[0013] The present invention also provides a virtual power plant configuration optimization system based on multi-agent collaborative game theory, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of the virtual power plant configuration optimization method based on multi-agent collaborative game theory described above.

[0014] The present invention has the following beneficial effects: This invention analyzes the differences in revenue between individual and collaborative operation of entities within a virtual power plant, assesses the collaborative value of each entity, quantifies the actual revenue enhancement brought by collaboration, and clarifies the expected value of their participation in the virtual power plant. Furthermore, it evaluates the dynamic response of each entity during actual configuration execution, using power fluctuation deviations during operation to characterize the degree of task undertaken, accurately reflecting the effort exerted by each entity. Combined with the entity's current power, the dynamic response level is determined, providing a quantitative basis for dynamically constructing a fair and effective revenue distribution mechanism. The invention analyzes the potential loss risks each entity may bear during virtual power plant operation, calculates risk compensation coefficients, improves the fairness and stability of the incentive mechanism, and prevents entities with strong resource capabilities from gradually withdrawing or becoming passive due to prolonged high loads. Combining risk compensation and dynamic response, the current contribution is determined, and based on the revenue matching between dynamic contribution capability and collaborative value, configuration decisions are optimized. This invention introduces a contribution-aware optimization scheme by matching the contributions and benefits of each entity in the collaborative process. While ensuring the maximization of overall benefits, it also takes into account the rationality of individual benefits and incentive compatibility, thereby improving the operating efficiency of the virtual power plant, the incentive effectiveness of benefit distribution, and the stability of the multi-entity alliance, and reducing the structural disconnect between configuration optimization and benefit distribution. Attached Figure Description

[0015] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 A flowchart illustrating a virtual power plant configuration optimization method based on multi-agent collaborative game theory, provided in one embodiment of the present invention; Figure 2 This is a schematic diagram of a virtual power plant configuration scheme provided in one embodiment of the present invention; Figure 3 This is a schematic diagram illustrating the distribution of actual power and expected power according to an embodiment of the present invention. Figure 4 This is a schematic diagram of the process of solving the optimized registration scheme using the ADMM algorithm system according to an embodiment of the present invention. Detailed Implementation

[0017] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a virtual power plant configuration optimization method and system based on multi-agent cooperative game theory proposed by the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0018] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0019] The following description, in conjunction with the accompanying drawings, details the specific scheme of the virtual power plant configuration optimization method and system provided by this invention, which involves multi-agent collaborative game theory.

[0020] Please see Figure 1 The diagram illustrates a flowchart of a virtual power plant configuration optimization method based on a multi-agent collaborative game theory according to an embodiment of the present invention. The method includes the following steps: S1: Obtain the collaborative benefits and actual power of each entity during each resource scheduling cycle in the collaborative runtime of the virtual power plant.

[0021] In a virtual power plant (VPP), a group of entities comprising various resources such as solar power, wind power, energy storage, and adjustable loads jointly participate in scheduling and service, achieving resource complementarity and increased revenue through collaborative operation. To achieve efficient resource access and optimized scheduling during multi-entity collaboration, a game theory model considering the differences among entities is needed to formulate a scientific resource allocation strategy at the system level. Please refer to [link / reference]. Figure 2 The diagram illustrates a flow chart of a virtual power plant configuration scheme provided by an embodiment of the present invention.

[0022] In this embodiment of the invention, the actual participation records of the subjects are retrieved from the virtual power plant platform. The key features include the expected power issued by the platform to them. The system formulates resource scheduling arrangements for each subject. The actual power of the subject is the active power that the subject actually injects into the system or absorbs from the system at a certain moment, measured in kilowatts (kW) by the smart meter in real time.

[0023] In this embodiment of the invention, the operating time of the virtual power plant over the past week is used as the collaborative operating period for analysis. When real-time optimization configuration decisions are made on a minute-by-minute basis, a 15-minute time unit is adopted as the resource scheduling cycle. Based on the fair allocation method of cooperative game theory, the Shapley value is used to obtain the collaborative benefit of each entity within a certain resource scheduling cycle. It should be noted that the implementer of the data collection settings can adjust them according to the specific implementation situation, and no restrictions are imposed here. The calculation of the Shapley value is a well-known technique familiar to those skilled in the art, and will not be described in detail here.

[0024] S2: Analyze the current collaborative value of an entity based on its independent operating benefits and collaborative benefits during the current resource scheduling cycle.

[0025] The value of collaboration reflects the value judgments and expected benefits of participating entities. If the value of collaboration deviates significantly from the benefits gained from individual operation, it can easily lead to decreased willingness to participate, reduced responsiveness, and even resource withdrawal and instability within the collaborative alliance. Therefore, it is necessary to reasonably assess individual benefit levels under the premise of maximizing overall system benefits, ensuring that the optimal allocation scheme possesses incentive compatibility and collaborative sustainability.

[0026] In a virtual power plant, each entity participates in collaboration to benefit from the alliance. The marginal benefits gained through resource sharing, unified scheduling, and complementary collaboration are greater than the benefits of operating independently. The greater the difference between the benefits allocated to an entity in collaborative operation and those in individual operation, the higher its collaborative value. The collaborative value of each entity is determined based on its benefits in collaborative and individual operation.

[0027] Preferably, in this embodiment of the invention, the method for obtaining collaborative value includes: First, the independent revenue of each entity operating independently within the current resource scheduling cycle is obtained. In this embodiment of the invention, historical operating data of the entity before joining the virtual power plant is obtained, including its power output or load absorption curves, autonomous scheduling strategies, and operating cost information. Combined with external factors such as meteorological conditions and electricity price curves, a dynamic programming (DP) model of the entity's autonomous operation "detached from the virtual power plant" is constructed. The model simulates and estimates the independent operating revenue of the entity during the same time period, serving as a reference for measuring the change in value before and after the entity participates in collaboration. It should be noted that the construction of the dynamic programming model is a technique well-known to those skilled in the art and will not be further elaborated upon here.

[0028] Then, the difference between the collaborative benefit and the independent benefit of each entity within the current resource scheduling cycle is normalized to obtain the current collaborative value of each entity. It should be noted that normalization is a technique well-known to those skilled in the art, and the choice of normalization function can be linear normalization or standard normalization, etc. The specific normalization method is not limited here.

[0029] S3: During the collaborative operation of a single entity, the fluctuation deviation coefficient of the entity is determined by the degree of change between the actual power fluctuation and the expected power supply deviation in consecutive adjacent resource scheduling cycles; based on the actual power fluctuation and expected power in the current resource scheduling cycle of a single entity, the current dynamic response of the entity is obtained by combining the fluctuation deviation coefficient.

[0030] In the collaborative operation of a virtual power plant, the actual effort required by each entity to execute configurations varies significantly due to differences in resource characteristics, responsiveness, and task division. For example, photovoltaic power may be unable to complete dispatch tasks due to output fluctuations, while energy storage or adjustable loads must bear additional responsibility for system balancing. Therefore, a comprehensive analysis of the dynamic response contributions of each entity can be conducted, and the revenue can be optimized in a coordinated manner based on the response contributions of each entity.

[0031] Virtual power plants aggregate various types of distributed resources, including fluctuating resources such as photovoltaic and wind power, which are significantly affected by the external environment, as well as relatively stable entities such as energy storage systems and adjustable loads. In actual operation, external factors such as extreme weather, equipment failures, and grid disturbances often prevent some entities from achieving their expected output according to the predetermined configuration, resulting in a mismatch between actual output and planned configuration. To improve the accuracy of resource scheduling and the feasibility of configuration schemes, based on historical operating data, the availability and responsiveness of various entity resources under external disturbance conditions are identified and quantified, providing a reliable basis for predicting the dynamic response level of each entity's resources within the current resource scheduling cycle.

[0032] By analyzing the fluctuations in actual power consumption and the temporal characteristics of deviations from planned power during historical collaborative operation periods, a fluctuation deviation coefficient is obtained for the subject, reflecting the relationship between actual resource consumption and deviations in the current historical time series. Preferably, in this embodiment of the invention, the method for obtaining the fluctuation deviation coefficient includes: For any given entity, its resource volatility in each resource scheduling cycle is obtained based on the degree of fluctuation and disorder of its actual power in each cycle, and the deviation from the actual power fluctuation in the resource scheduling cycle during the collaborative operation phase. During the collaborative operation phase of the virtual power plant, the actual power output of each entity to the system in each resource scheduling cycle is analyzed. If the actual power change within a cycle is large and exceeds the normal level, it may encounter external interference, and thus the resource volatility of the entity in that cycle will be greater.

[0033] In this embodiment of the invention, the variance of the actual power of the subject in each resource scheduling cycle is used as the degree of disorder variation in each resource scheduling cycle, reflecting the degree of power variation within each cycle. The mean of the degree of disorder variation of the subject in all resource scheduling cycles during the cooperative runtime is used as the mean of the subject's variation, representing the general degree of power variation of the subject in recent cooperative runtimes.

[0034] Then, the difference between the degree of disorder and the mean change in each resource scheduling cycle is calculated. The ratio of the difference to the mean change is used as the fluctuation deviation in each resource scheduling cycle, quantifying the degree to which the power change in a single resource scheduling cycle deviates from the general level. Finally, the product of the fluctuation deviation and the degree of disorder in each resource scheduling cycle is normalized to obtain the resource volatility of the entity.

[0035] Furthermore, the difference between the actual power and the expected power of the entity at each moment in each resource scheduling cycle is calculated. The mean of all differences is normalized to obtain the resource supply variability of the entity in each resource scheduling cycle. By comparing and analyzing the actual output power with the expected output power in each resource scheduling cycle, the scheduling uncertainty caused by power fluctuations can be measured. Please refer to [link / reference]. Figure 3 The diagram illustrates the distribution of actual power and expected power according to an embodiment of the present invention. Curve A represents the expected power required by the virtual power plant scheduling resource command, and curve B represents the actual power. In subsequent scheduling, due to certain factors, the power output of the resources provided by the entity deviates significantly.

[0036] Analyzing the relationship between resource volatility and resource supply variability in the recent operation of each entity helps predict the current possible output power deviation. If the magnitude of the change in resource supply variability and the magnitude of resource volatility tend to be consistent in adjacent cycles, it indicates that resource volatility can be used to infer the dynamic response of the entity in the current cycle.

[0037] Therefore, during the collaborative operation period of this entity, the ratio of the difference in resource volatility to the difference in resource supply between two adjacent resource scheduling cycles is taken as the adjacent change ratio between two adjacent resource scheduling cycles, reflecting the proportional relationship of relative change.

[0038] By combining the proportional relationships across all consecutive cycles, the average of the adjacent change ratios of all two adjacent resource scheduling cycles during the collaborative runtime is used as the fluctuation deviation coefficient of the subject. Based on the fluctuation deviation coefficient, the actual deviation of the subject's expenditure in the current resource scheduling cycle can be predicted and analyzed.

[0039] When each entity receives a resource scheduling task from the virtual power plant, the actual output may deviate due to volatility. The deviation between the expected power and the predicted power under the volatility is analyzed. The greater the volatility, the greater the deviation in resource supply, and the smaller the actual dynamic response of the entity.

[0040] Preferably, in this embodiment of the invention, the method for obtaining dynamic response includes: For any given entity, the actual power variance of that entity within the current resource scheduling cycle is taken as the entity's current power volatility. The product of the entity's current volatility and the volatility deviation coefficient is taken as the entity's current response deviation. By observing the overall deviation change pattern over historical time series, the expected deviation of the entity's current actual expenditure is obtained.

[0041] The difference between the expected power of the subject in the current resource scheduling cycle and the current response deviation is further normalized to obtain the current dynamic response of the subject. By referring to the influence of historical scheduling, the current dynamic response is estimated.

[0042] S4: Based on the equipment loss of a single entity's equipment in the current resource scheduling cycle and its relationship with other entities, analyze the current risk compensation coefficient of the single entity.

[0043] When entities in a virtual power plant need to provide more resources, the risks, losses, or maintenance costs borne by each entity also increase. If the losses are greater compared to other entities, then more risk compensation benefits need to be provided. Therefore, based on the resource scheduling arrangement for each entity through the configuration scheme in the current cycle, that is, the losses of each entity under the expected power output in the current resource scheduling cycle, the risk compensation coefficient of each entity is obtained.

[0044] In this embodiment of the invention, the equipment loss of each subject under the expected power in the current resource scheduling cycle is obtained, and the difference between the equipment loss of each subject and the average loss of all subject equipment is normalized to obtain the risk compensation coefficient of each subject. The larger the risk compensation coefficient, the higher the risk that the subject bears in providing resources, and the higher it should be in the subsequent contribution capacity assessment.

[0045] In one specific embodiment of the present invention, the device loss under the expected power can be obtained from the attenuation model provided by the corresponding manufacturer of the main body, which will not be elaborated here.

[0046] S5: Analyze the current dynamic contribution capability of an entity by using its current dynamic responsiveness and risk compensation coefficient; and obtain the contribution-return matching degree of an entity based on the matching relationship between its current dynamic contribution capability and collaborative value.

[0047] Since the higher the risk compensation coefficient, the more risk the entity bears in providing resources, the greater its actual dynamic contribution should be. Entities that primarily provide resources in the long term will contribute more. Therefore, the dynamic contribution capability can be obtained by analyzing the entity's current dynamic responsiveness and risk compensation coefficient.

[0048] In this embodiment of the invention, for any given subject, the product of the subject's risk compensation coefficient and dynamic responsiveness is taken as the subject's risk adjustment degree, which is also taken as the amplification adjustment degree. The sum of the subject's risk adjustment degree and dynamic responsiveness is taken as the subject's dynamic contribution capability.

[0049] By comprehensively analyzing the dynamic contribution behavior and collaborative value of each entity, a direct correspondence can be established between the entity's actual input and the benefits it receives. This enables the linkage optimization of resource allocation decisions and benefit distribution mechanisms, ensuring that the virtual power plant maximizes the overall system benefits while also guaranteeing the rationality of each entity's benefits and the fairness of incentives. This, in turn, improves the overall collaborative efficiency and the continuous stability of system operation.

[0050] When entities collaborate in the operation of a virtual power plant, the collaboration value represents the entity's benefit, since the purpose of each entity's participation in the operation of the virtual power plant is to achieve this. The greater the entity's dynamic contribution capability in the current cycle, the higher the collaboration value, and the more reasonable the entity's benefit, and the greater the degree of matching between contribution and benefit.

[0051] In this embodiment of the invention, the difference between a single entity's current dynamic contribution capability and collaborative value is negatively correlated and normalized to obtain the entity's contribution-reward matching degree. The greater the contribution-reward matching degree, the greater the degree of matching between the entity's contribution capability and its rewards. It should be noted that negative correlation mapping is a technique well-known to those skilled in the art, such as using inverse proportional or negative exponential forms, etc., and will not be elaborated or limited here.

[0052] As an example, the expression for contribution-reward matching degree is: In the formula, Represented as the first The degree of matching between the current contribution and benefits of each entity; Represented as the first The collaborative value of each entity in the current resource scheduling cycle; Represented as the first The dynamic contribution capability of each entity in the current resource scheduling cycle. Represented as an absolute value extraction function, It is represented as an exponential function with the natural constant as the base.

[0053] S6: Obtain a configuration scheme through distributed optimization based on the matching degree of contribution and benefit.

[0054] In the collaborative operation of a virtual power plant involving multiple stakeholders, the core objective of configuration optimization is to maximize the overall benefits of the system, which may overlook the actual interests of individual participants. To prevent free-riding, incentive failure, and resource idleness, it is necessary to ensure that the input and output of each participant are matched, maintain the stability of the alliance, and cooperate with distributed optimization algorithms to achieve a collaborative solution method of "computing and negotiating simultaneously". The benefit distribution mechanism is introduced into the configuration optimization problem as part of the solution.

[0055] The configuration optimization model for virtual power plants needs to comprehensively consider the economic benefits at the system level. That is, under the premise of satisfying power balance, equipment constraints, and market rules, it should maximize the overall operating revenue, such as the revenue from electricity generation minus the operating costs. At the same time, it should achieve fair distribution among all stakeholders, that is, ensure that there is a high degree of matching between the collaborative contributions made by each participating entity and the revenue it receives.

[0056] Therefore, in this embodiment of the invention, the contribution-benefit matching degree of each entity is used as a weight, and the optimal configuration scheme is obtained through collaborative solution using the ADMM algorithm. In a virtual power plant with multiple entities, to achieve efficient optimization and fair cooperation in resource allocation, a distributed optimization algorithm, such as the Alternating Direction Multiplier Method (ADMM algorithm), is adopted. The contribution-benefit matching degree of each entity is introduced into the local objective function, constructing a collaborative solution mechanism of "computing and negotiating simultaneously" to obtain the optimal configuration scheme. The goal is to maximize the overall benefit of the virtual power plant, reduce the cost of each entity, and maximize its benefit. Please refer to [link to relevant documentation]. Figure 4 The diagram illustrates a flowchart of an ADMM algorithm system for solving an optimized registration scheme, according to an embodiment of the present invention.

[0057] It should be noted that the method of solving cooperative coordination tasks among multiple agents through distributed optimization algorithms is a well-known technique familiar to those skilled in the art, and will not be elaborated here.

[0058] In one specific embodiment of the present invention, the process of obtaining the optimized configuration scheme includes: first, decomposing the resource allocation optimization problem of the entire virtual power plant into multiple local sub-problems, with each resource entity independently solving its own optimal strategy. Second, incorporating the contribution-benefit matching degree of each entity into its local objective function, guiding the entity's optimization direction as a weighting factor. Then, in each iteration, each entity performs optimization locally, solves its own objective function, and reports the results (such as power allocation, response capability, etc.). Finally, the virtual power plant system platform aggregates the results of all entities, calculates globally consistent variables (such as total system scheduling or price, etc.), and broadcasts them to each entity. Each entity updates its multiplier variables based on the global feedback information and adjusts its optimization direction for the next round.

[0059] Ultimately, the resource optimization scheme, during its implementation, not only aligns with the overall objective of maximizing the total revenue of the virtual power plant, but also fully considers the dynamic contributions, collaborative value, and fair benefits of each entity, thus forming a robust, efficient, and incentive-compatible collaborative allocation scheme.

[0060] In summary, this invention analyzes the differences in revenue between individual and collaborative operation of each entity in a virtual power plant, assesses the collaborative value of individual entities, quantifies the actual revenue improvement brought by collaboration to each entity, and clarifies the expected value of their participation in the virtual power plant. Furthermore, it evaluates the dynamic response of each entity in actual configuration execution, using power fluctuation deviations during actual operation to characterize the degree of task undertaken, accurately reflecting the effort exerted by each entity. Combined with the entity's current power, the dynamic response degree is determined, providing a quantitative basis for dynamically constructing a fair and effective revenue distribution mechanism. The invention analyzes the potential loss risks each entity may bear in the operation of the virtual power plant, calculates risk compensation coefficients, improves the fairness and stability of the incentive mechanism, and prevents entities with strong resource capabilities from gradually withdrawing or becoming passive due to prolonged high loads. Combining risk compensation and dynamic response, the current contribution is determined, and based on the revenue matching between dynamic contribution capability and collaborative value, configuration decisions are optimized. This invention introduces a contribution-aware optimization scheme by matching the contributions and benefits of each entity in the collaborative process. While ensuring the maximization of overall benefits, it also takes into account the rationality of individual benefits and incentive compatibility, thereby improving the operating efficiency of the virtual power plant, the incentive effectiveness of benefit distribution, and the stability of the multi-entity alliance, and reducing the structural disconnect between configuration optimization and benefit distribution.

[0061] The present invention also provides a virtual power plant configuration optimization system based on multi-agent collaborative game theory, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of the virtual power plant configuration optimization method based on multi-agent collaborative game theory described above.

[0062] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0063] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A method for optimizing the configuration of a virtual power plant through multi-agent collaborative game theory, characterized in that, The method includes: Obtain the collaborative benefits and actual power of each entity during each resource scheduling cycle in the collaborative runtime of the virtual power plant; Based on the independent operating benefits and collaborative benefits of a single entity in the current resource scheduling cycle, analyze the current collaborative value of the entity; During the collaborative operation of a single entity, the fluctuation deviation coefficient of the entity is determined by the degree of change between the actual power fluctuation and the expected power supply deviation in consecutive adjacent resource scheduling cycles. Based on the actual power fluctuation and expected power in the current resource scheduling cycle of a single entity, the current dynamic responsiveness of the entity is obtained by combining the fluctuation deviation coefficient. Based on the equipment loss of a single entity's equipment in the current resource scheduling cycle and its relationship with other entities, analyze the current risk compensation coefficient of the single entity; By analyzing the current dynamic responsiveness and risk compensation coefficient of the subject, we can determine the current dynamic contribution capability of the subject; based on the matching relationship between the current dynamic contribution capability and the collaborative value of a single subject, we can obtain the contribution-return matching degree of the subject. Configuration schemes are obtained through distributed optimization based on the matching degree of contribution and benefit.

2. The method for optimizing the configuration of a virtual power plant through multi-agent collaborative game theory as described in claim 1, characterized in that, The methods for obtaining the value of collaboration include: Obtain the independent revenue of each entity operating independently within the current resource scheduling cycle; The difference between the collaborative gains and independent gains of each entity within the current resource scheduling cycle is normalized to obtain the current collaborative value of each entity.

3. The method for optimizing the configuration of a virtual power plant through multi-agent collaborative game theory as described in claim 1, characterized in that, The method for obtaining the fluctuation deviation coefficient includes: For any entity, the resource volatility of the entity in each resource scheduling cycle is obtained based on the degree of fluctuation and disorder of the actual power of the entity in each resource scheduling cycle and the deviation of the actual power fluctuation from the resource scheduling cycle in the cooperative runtime. Calculate the difference between the actual power and the expected power of the entity at each moment in each resource scheduling cycle, and normalize the mean of all differences to obtain the resource supply variability of the entity in each resource scheduling cycle. During the collaborative operation period of this entity, the ratio of the difference in resource volatility to the difference in resource supply between any two adjacent resource scheduling cycles is taken as the adjacent change ratio between any two adjacent resource scheduling cycles; the average of the adjacent change ratios between all two adjacent resource scheduling cycles during the collaborative operation period is taken as the volatility deviation coefficient of this entity.

4. The method for optimizing the configuration of a virtual power plant through multi-agent collaborative game theory as described in claim 3, characterized in that, The method for obtaining resource volatility includes: The variance of the actual power of the subject in each resource scheduling cycle is taken as the degree of disorder variation in each resource scheduling cycle; the mean of the degree of disorder variation of the subject in all resource scheduling cycles during the cooperative runtime is taken as the mean of the subject's variation. Calculate the difference between the degree of disorder and the mean of change for each resource scheduling cycle, and use the ratio of the difference to the mean of change as the degree of fluctuation deviation for each resource scheduling cycle. The resource volatility of the entity is obtained by normalizing the product of the fluctuation deviation and the disorder variation in each resource scheduling cycle.

5. The method for optimizing the configuration of a virtual power plant through multi-agent collaborative game theory as described in claim 1, characterized in that, The method for obtaining the dynamic response includes: For any subject, the actual power variance of the subject within the current resource scheduling cycle is taken as the current power volatility of the subject; the product of the subject's current volatility and volatility deviation coefficient is taken as the subject's current response deviation. The difference between the expected power of the subject in the current resource scheduling cycle and the current response deviation is normalized to obtain the current dynamic response of the subject.

6. The method for optimizing the configuration of a virtual power plant through multi-agent collaborative game theory as described in claim 1, characterized in that, The method for obtaining the risk compensation coefficient includes: Obtain the equipment loss of each entity under the expected power in the current resource scheduling cycle; The difference between the equipment loss of each entity and the average loss of all entities' equipment is normalized to obtain the risk compensation coefficient for each entity.

7. The method for optimizing the configuration of a virtual power plant through multi-agent collaborative game theory as described in claim 1, characterized in that, The methods for obtaining the dynamic contribution capability include: For any given entity, the product of its risk compensation coefficient and dynamic responsiveness is taken as its risk adjustment degree; the sum of its risk adjustment degree and dynamic responsiveness is taken as its dynamic contribution capability.

8. The method for optimizing the configuration of a virtual power plant through multi-agent collaborative game theory as described in claim 1, characterized in that, The methods for obtaining the contribution-benefit matching degree include: By negatively mapping and normalizing the difference between the current dynamic contribution capacity and collaborative value of a single entity, the contribution-benefit matching degree of the entity is obtained.

9. The method for optimizing the configuration of a virtual power plant through multi-agent collaborative game theory as described in claim 1, characterized in that, The configuration scheme obtained through distributed optimization based on contribution-benefit matching degree includes: The optimal configuration scheme is obtained by using the matching degree of contribution and benefit of each subject as a weight and through collaborative solution using the ADMM algorithm.

10. A virtual power plant configuration optimization system based on multi-agent collaborative game theory, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the virtual power plant configuration optimization method for multi-agent collaborative game as described in any one of claims 1 to 9.

Images