Search

CN-122026522-A - Virtual power plant response scheduling method and device based on dynamic Bayesian game excitation

CN122026522ACN 122026522 ACN122026522 ACN 122026522ACN-122026522-A

Abstract

The invention discloses a virtual power plant response scheduling method and device based on dynamic Bayesian game excitation, wherein the method comprises the steps of firstly constructing a three-layer game framework taking a power grid as a leading person, an aggregation provider and a user as a following person; then respectively constructing a first objective function of a power grid, a second objective function of an aggregator and a third objective function of a user, and constructing a game model based on the three-layer game framework and the corresponding objective functions; and finally, solving the game model by adopting a reverse induction method to obtain a virtual power plant response scheduling result. The posterior beliefs are obtained by carrying out weighted fusion on rational Bayesian beliefs, private anchor point beliefs of aggregators and public anchor point beliefs of the power grid in the third objective function, and a foreground theory is introduced, so that fine and complex descriptions of user behaviors are constructed, and the existing price-quantity game is converted into a dynamic game between the power grid and the aggregators, wherein the dynamic game contends for the leading rights of the users, thereby improving the accuracy of the blog.

Inventors

  • XU WENYUAN
  • TAO YUAN
  • XU JI

Assignees

  • 湖北华中电力科技开发有限责任公司

Dates

Publication Date
20260512
Application Date
20260408

Claims (10)

  1. 1. A virtual power plant response scheduling method based on dynamic Bayesian game excitation is characterized by comprising the following steps: constructing a three-layer game framework taking a power grid as a master, an aggregator and a user as a follower; According to the current social value of the power grid and the maximum expected discount social value which can be realized by the power grid based on the current system state, a first objective function of the power grid is constructed with the aim of maximizing the total expected profit of the power grid; Constructing a second objective function of the aggregator with the aim of maximizing the total expected profit of the aggregator according to the current profit of the aggregator and the maximum expected discount social value which can be realized by the aggregator based on the current system state; Constructing a third objective function of the user by taking the maximized foreground theoretical value as a target according to posterior beliefs, private type parameters of the user, unit load reduction compensation price provided by the aggregator to the user and additional service value of the aggregator, wherein the posterior beliefs are obtained by carrying out Softmax weighted fusion on rational Bayesian beliefs, private anchor point beliefs of the aggregator and public anchor point beliefs of the power grid; Constructing a game model based on the three-layer game framework, the first objective function, the second objective function and the third objective function; and solving the game model by adopting a reverse induction method to obtain a virtual power plant response scheduling result.
  2. 2. The method for scheduling virtual power plant response based on dynamic bayesian game excitation according to claim 1, wherein constructing a first objective function of the power grid with the objective of maximizing a total expected profit of the power grid according to a current social value of the power grid and a maximum expected discount social value that the power grid can achieve based on a current system state comprises: Obtaining a current social value according to the social value obtained by the system due to the reduction of the actual load of a user and the total cost required to be paid for obtaining the reduction of the load; Calculating the maximum expected discount social value of the power grid based on the current system state according to the first discount factor and the maximum expected discount social value function under the current system state; And obtaining a first objective function according to the current social value and the maximum expected discount social value which can be realized by the power grid based on the current system state.
  3. 3. The method for scheduling virtual power plant response based on dynamic bayesian gaming excitation according to claim 2, wherein the first objective function is: Wherein, the Indicating that the power grid is at the first Each scheduling period, Under the condition, the maximum expectations that can be realized compromise the social cost function, as the objective function, Is the first A system state vector for each scheduling period including user belief anchor, trust strength, historical price, and aggregator belief of user type, For the power grid The settlement price set in the scheduling period, On the first place for the electric network The price of the guideline issued during a scheduling period, As a private-type parameter of the aggregator, In order for the operator to be desirable for the condition, As a social cost function of the power grid in the current scheduling period, As a result of the first fold factor, For the desired operator for system state transition uncertainty, For the next periodic system state vector formed after the current periodic decision and user response are completed, Indicating that the power grid is at the first Each scheduling period, Under the condition, the maximum expectations that can be realized compromise the social cost function.
  4. 4. The method for scheduling virtual power plant response based on dynamic bayesian game excitation according to claim 1, wherein the second objective function is: Wherein, the Is based on The desire for the type of user is that, For the current profit to be a matter of interest, For the aggregator at the first Within a scheduling period, And self-type The optimal desired discount cost function, i.e. the second objective function, Is the first A system state vector for each scheduling period including user belief anchor, trust strength, historical price information, and the aggregator's belief of user type, As a private-type parameter of the aggregator, For the aggregator at the first The unit load released to the user during a scheduling period cuts off the offset price, For the aggregator at the first Posterior beliefs for user types over a period, As a second of the fold factors, the first fold factor, For the purpose of a future function of value, For the desired operator of uncertainty in the state transition of the system, To at the first And after the end of the cycle decision and the user response, the system evolves to obtain the next cycle state vector.
  5. 5. The virtual power plant response scheduling method based on dynamic bayesian game excitation according to claim 1, wherein the third objective function is: Wherein, the The decision variable is decremented for the load selected by the user during the current scheduling period, As a user utility function based on foreground theory, For the user at the first The final posterior beliefs formed during the scheduling periods, As a private-type parameter of the user, For the aggregator at the first The unit load offered to the user during a scheduling period cuts off the compensation price, As a revenue cost function in the foreground theory for describing the nonlinear value perception of forward revenue by a user within the revenue domain, As a loss cost function in the foreground theory that describes the nonlinear value perception of negative returns by users in the return domain, For the aggregate to have an additional service cost function, Is private type parameter of aggregator for distinguishing efficient aggregator from inefficient aggregator Based on posterior beliefs of users The calculated expected value of the aggregate extra service value.
  6. 6. The virtual power plant response scheduling method based on dynamic Bayesian game excitation as claimed in claim 3, wherein the posterior beliefs are calculated in the following manner: Wherein, the The information source is represented by a reference to the information source, , Representing the rational bayesian source of information, Representing the private anchor source of the aggregator, Representing a source of a common anchor point of the power grid, Representing a source of information In the first place The dynamic weights of the individual scheduling periods are, Representing a source of information In the first place The beliefs of the individual scheduling periods, dynamic weights are derived from trust strength and evidence titers.
  7. 7. The method for scheduling virtual power plant response based on dynamic bayesian gaming excitation according to claim 6, wherein when solving the gaming model by using a reverse induction method, the method further comprises: Updating the dynamic weight according to the remorse value of the counterfactual, wherein the remorse value of the counterfactual is obtained by calculating the theoretical utility value of the counterfactual foreground and the maximum value in all counterfactual utilities, which can be obtained by a user when the user makes a decision according to the beliefs of the information sources, and the magnitude of the counterfactual remorse value is used for representing the post-instruction effect of the corresponding information sources on the user decision; and updating the trust strength by adopting a memory decay mechanism based on the remorse value of the counterfactual.
  8. 8. A virtual power plant response scheduling device based on dynamic bayesian game excitation, comprising: The game framework construction module is used for constructing a three-layer game framework taking a power grid as a master, an aggregator and a user as a follower; The first objective function construction module is used for constructing a first objective function of the power grid by taking the total expected profit of the power grid as the target of maximizing the current social value of the power grid and the maximum expected discount social value which can be realized by the power grid based on the current system state; The second objective function construction module is used for constructing a second objective function of the aggregator by taking the total expected profit of the aggregator as the target of maximizing the total expected profit of the aggregator according to the current profit of the aggregator and the maximum expected discount social value which can be realized by the aggregator based on the current system state; the third objective function construction module is used for constructing a third objective function of the user by taking the maximum foreground theoretical value as a target according to posterior beliefs, private type parameters of the user, unit load reduction compensation price provided by an aggregator to the user and additional service value of the aggregator, wherein the posterior beliefs are obtained by carrying out Softmax weighted fusion on rational Bayesian beliefs, private anchor point beliefs of the aggregator and public anchor point beliefs of a power grid; The game model building module is used for building a game model based on the three-layer game framework, the first objective function, the second objective function and the third objective function; and the model solving module is used for solving the game model by adopting an inverse induction method to obtain a virtual power plant response scheduling result.
  9. 9. A computer readable storage medium, having stored thereon a computer program which when executed by a processor implements the dynamic bayesian game excitation based virtual power plant response scheduling method of any of claims 1 to 7.
  10. 10. A computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the dynamic bayesian game excitation-based virtual power plant response scheduling method of any of claims 1 to 7 when the program is executed by the processor.

Description

Virtual power plant response scheduling method and device based on dynamic Bayesian game excitation Technical Field The invention relates to the technical field of smart grids, in particular to a virtual power plant response scheduling method and device based on dynamic Bayesian game excitation. Background The virtual power plant is a set of software platform system with digital technology, control technology, internet of things technology and information communication technology, and the core of the virtual power plant is communication and aggregation. The virtual power plant response scheduling means that scattered distributed energy resources (such as photovoltaics, wind power, energy storage, adjustable loads, electric vehicles and the like) are aggregated into a virtual whole which can be uniformly scheduled through advanced communication, control and optimization algorithms, and the output or load behaviors of the virtual power plant response scheduling means are dynamically adjusted according to the running requirements of a power grid or electric market signals, so that the functions of peak clipping, valley filling, frequency modulation, voltage regulation, auxiliary service and the like are realized. The prior art adopts a three-layer demand response architecture comprising a power grid, an aggregator and users, wherein the operation cost of the aggregator and the comfort preference of the users are private information, and the asymmetry of the information is a key problem affecting the market efficiency. In order to solve the above problems, a three-layer bayesian-stark-berg game model is proposed in some prior art, in which a top-layer grid G is used as a master to set a settlement price, a middle-layer aggregator a is used as a follower to select a compensation price after observing the settlement price, and a bottom-layer user U selects an optimal load reduction amount after observing the compensation price. The core mechanism of the scheme is that the compensation price of the aggregator is not only economic compensation, but also a signal. After observing the compensation price, the user U is taken as a signal receiver, updates posterior beliefs of the user U to the type of the aggregator by using a Bayesian rule, and designs a user utility function based on the posterior beliefs, so that the technical support or service value provided by the efficient aggregator is effectively utilized. However, the above-mentioned technology still has the following technical problems that, firstly, the existing model assumes a completely rational bayesian, the beliefs can be strictly updated according to bayesian rule, and the expected utility is strictly maximized, which is not in accordance with reality, a large number of researches prove that individuals have unexpected utility characteristics such as loss aversion, reference point dependence and the like in decision making, and have cognitive bias errors such as anchoring effect and the like in the course of belief updating, so that the calculation of posterior beliefs is not in accordance with reality, in addition, the existing model adopts static games, the interaction relationship among a power grid, an aggregator and users is dynamically changed, and the model cannot accurately describe or describe the relationship, so that the response scheduling effect is poor. Disclosure of Invention The invention aims to provide a virtual power plant response scheduling method and device based on dynamic Bayesian game excitation, which are used for solving the technical problem of poor response scheduling effect caused by inconsistent posterior belief calculation and static game framework in the prior art. In order to achieve the above object, a first aspect of the present invention provides a virtual power plant response scheduling method based on dynamic bayesian game excitation, including: constructing a three-layer game framework taking a power grid as a master, an aggregator and a user as a follower; According to the current social value of the power grid and the maximum expected discount social value which can be realized by the power grid based on the current system state, a first objective function of the power grid is constructed with the aim of maximizing the total expected profit of the power grid; Constructing a second objective function of the aggregator with the aim of maximizing the total expected profit of the aggregator according to the current profit of the aggregator and the maximum expected discount social value which can be realized by the aggregator based on the current system state; Constructing a third objective function of the user by taking the maximized foreground theoretical value as a target according to posterior beliefs, private type parameters of the user, unit load reduction compensation price provided by the aggregator to the user and additional service value of the aggregator, wherein the posterior beliefs are obtained by carr