CN-121984111-A - Micro-grid scheduling method, device, equipment and medium based on dynamic weight game

CN121984111ACN 121984111 ACN121984111 ACN 121984111ACN-121984111-A

Abstract

The invention discloses a micro-grid dispatching method, device, equipment and medium based on dynamic weight game, wherein the method comprises the steps of constructing and solving a non-cooperative game model to obtain balanced benefits by collecting real-time operation data and identifying a multi-benefit main body; generating a dynamic weight vector based on balanced income, constructing an approximate dynamic planning scheduling model by combining the dynamic weight and updated data, outputting scheduling instructions in two stages of daily optimization and daily rolling optimization and updating model parameters, and finally feeding back corrected weight according to actual running deviation to form closed-loop optimization. The invention effectively improves the accuracy of micro-grid dispatching and reduces the comprehensive operation cost.

Inventors

HU DONGBIN
YU QINGHAO
QIN KE
YANG SHUHAN
ZHANG JUNHAO

Assignees

湖南红普创新科技发展有限公司

Dates

Publication Date: 20260505
Application Date: 20251223

Claims (10)

1. A micro-grid scheduling method based on dynamic weight game is characterized by comprising the following steps: collecting real-time running data of a micro-grid, and identifying an operator, a distributed power source owner, an energy storage operator and a user in the micro-grid as game main bodies; Based on the real-time operation data and the identified game main bodies, constructing a non-cooperative dynamic game model and solving to obtain balanced benefits of each game main body; Based on the balanced income, calculating initial weights of three targets of economy, low carbon and reliability, and carrying out optimized search by taking the initial weights as the center to generate dynamic weight vectors; Acquiring updated operation data, constructing a system state space and a scheduling action space based on the dynamic weight vector and the updated operation data, defining a multi-objective instant cost function and an approximate cost function, and forming an approximate dynamic planning scheduling model; Based on the approximate dynamic programming scheduling model, performing daily scheduling optimization and daily rolling scheduling optimization, outputting scheduling instructions, and updating parameters of the approximate cost function in the daily rolling scheduling optimization process; And after the scheduling instruction is executed, acquiring the actual running cost, calculating the deviation between the actual running cost and the expected target cost, generating a weight correction amount when the deviation exceeds a set threshold value, and applying the weight correction amount to the generation process of the dynamic weight vector of the next round.
2. The method of claim 1 wherein said constructing and solving a non-cooperative dynamic gaming model to obtain an equilibrium benefit for each of said gaming subjects comprises: Constructing a comprehensive utility function containing economic benefits, carbon emission costs and reliability risks for each game principal; Establishing a non-cooperative game model based on the comprehensive utility functions of all the game main bodies; and solving the non-cooperative game model by adopting Nash equilibrium or Stackelberg equilibrium to obtain the equilibrium benefits.
3. The method of claim 1, wherein calculating initial weights for three objectives of economy, low carbon, and reliability based on the balanced benefit comprises: Calculating contribution degrees of the game main bodies to the three targets of economy, low carbon and reliability according to the balanced benefits; defining a coalition function of the cooperative game based on the contribution; calculating the alliance function by using a Shapley value formula to obtain Shapley values of the three targets; and carrying out normalization processing on the Shapley values of the three targets to obtain the initial weight.
4. The method of claim 3, wherein the performing an optimization search centered on the initial weights generates dynamic weight vectors, comprising: constructing an optimization problem targeting a weighted sum of economic satisfaction, low-carbon satisfaction and reliability satisfaction with the initial weight as a search starting point; solving the optimization problem by adopting a particle swarm optimization algorithm to obtain an optimization weight; And carrying out weighted fusion on the optimized weight and the initial weight through a preset fusion coefficient to generate the dynamic weight vector.
5. The method of claim 1, wherein the building a system state space and a scheduling action space comprises: The system state space consists of a renewable energy output predicted value, a load predicted value, an energy storage state of charge, grid-connected electricity price and a system risk index; The scheduling action space is composed of an energy storage charging and discharging power instruction, a distributed power output instruction and a demand response adjustment quantity.
6. The method of claim 1, wherein updating parameters of the approximate cost function during the intra-day rolling schedule optimization process comprises: Executing a scheduling instruction in a rolling time domain, and collecting generated system state transition data and corresponding actual time value; Calculating a cost function error based on the system state transition data and the actual instant cost value by using a time sequence difference algorithm; and based on the cost function error, updating parameters of the approximate cost function by adopting a gradient descent method.
7. The method of claim 1, wherein generating a weight correction when the deviation exceeds a set threshold comprises: counting the average value of the actual running cost in a preset feedback period, and calculating the difference value between the average value and the expected target cost; Converting the difference value into adjustment amounts of three target weights of economy, low carbon and reliability according to a preset mapping rule; and performing low-pass filtering processing on the adjustment quantity to obtain the weight correction quantity.
8. A dynamic weight gaming-based micro-grid scheduling device, comprising: the data acquisition module is used for acquiring real-time operation data of the micro-grid and identifying an operator, a distributed power source owner, an energy storage operator and a user in the micro-grid as game main bodies; The profit solving module is used for constructing a non-cooperative dynamic game model based on the real-time operation data and the identified game main bodies and solving to obtain balanced profits of the game main bodies; the weight generation module is used for calculating initial weights of three targets of economy, low carbon and reliability based on the balanced benefits, and carrying out optimized search by taking the initial weights as the center to generate dynamic weight vectors; the model construction module is used for acquiring updated operation data, constructing a system state space and a scheduling action space based on the dynamic weight vector and the updated operation data, defining a multi-target instant cost function and an approximate cost function, and forming an approximate dynamic planning scheduling model; The model execution module is used for executing daily scheduling optimization and daily rolling scheduling optimization based on the approximate dynamic programming scheduling model, outputting scheduling instructions and updating parameters of the approximate cost function in the daily rolling scheduling optimization process; And the weight updating module is used for acquiring the actual running cost after executing the scheduling instruction, calculating the deviation between the actual running cost and the expected target cost, generating a weight correction amount when the deviation exceeds a set threshold value, and applying the weight correction amount to the generation process of the dynamic weight vector of the next round.
9. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the dynamic weight game based microgrid scheduling method of any one of claims 1 to 7 when the computer program is executed.
10. A computer readable storage medium storing a computer program, wherein the computer program when executed by a processor implements the dynamic weight game based microgrid scheduling method of any one of claims 1 to 7.

Description

Micro-grid scheduling method, device, equipment and medium based on dynamic weight game Technical Field The invention relates to the technical field of power dispatching, in particular to a micro-grid dispatching method, device, equipment and medium based on dynamic weight game. Background Along with the continuous improvement of the permeability of renewable energy sources such as photovoltaic, wind power and the like in a micro-grid, the power generation output of the renewable energy sources is influenced by natural conditions and shows remarkable randomness and fluctuation. Meanwhile, the application of the demand side response technology enables the user load to be actively adjusted along with the electricity price signal, so that the uncertainty of the load side is further increased. In this context, microgrid operation presents a complex feature of "energy-load" double-sided stochastic superposition. In order to ensure stable, economical and low-carbon operation of the system, a scheduling method capable of adapting to dynamic environment changes needs to be constructed, various distributed resources are coordinated and optimized, and a micro-grid scheduling technology becomes a core focus of the field. Currently, existing methods in the micro-grid dispatching field mainly have the following limitations. First, most scheduling models use fixed weights or static weights based on experience to weight sum different targets (e.g., economy, low carbon, reliability) on target weight settings. The static setting is difficult to flexibly adjust according to dynamic working conditions such as real-time output fluctuation of new energy, actual state of charge of stored energy, change of market price and the like, and deviation between a scheduling strategy and actual operation requirements is easy to occur in a scene of high-proportion renewable energy access. Second, in terms of benefit coordination, the actual operation of the microgrid involves multiple independent benefit principals of operators, distributed power owners, energy storage operators, users, etc., each principal goal having a natural conflict. The existing method is often optimized from a single-main-body view or approximates to the multi-objective problem by manually setting compromise weights, and the lack of effective depiction of strategy interaction and benefit gaming mechanisms between the multi-main-bodies makes a theoretical optimal scheduling scheme difficult in actual landing of multi-party cooperation. Finally, at the aspect of an optimization algorithm, the problems of complex modeling and low solving speed usually exist when the traditional methods such as random planning, robust optimization and the like face a high-dimensional state space, and the real-time requirement of online scheduling is difficult to meet. In summary, the existing micro-grid scheduling technology cannot effectively solve the problems of the target weight static solidification, difficulty in multi-body benefit coordination, insufficient dynamic adaptability of an optimization algorithm and the like, which are related to each other, so that the problems of poor real-time performance and inaccurate scheduling occur in the scheduling process of various distributed resources. Disclosure of Invention The embodiment of the invention provides a micro-grid dispatching method, a micro-grid dispatching device, computer equipment and a storage medium based on dynamic weight game, so as to improve the accuracy and timeliness of micro-grid dispatching. In order to solve the above technical problems, an embodiment of the present application provides a micro-grid scheduling method based on dynamic weight game, including: collecting real-time running data of a micro-grid, and identifying an operator, a distributed power source owner, an energy storage operator and a user in the micro-grid as game main bodies; Based on the real-time operation data and the identified game main bodies, constructing a non-cooperative dynamic game model and solving to obtain balanced benefits of each game main body; Based on the balanced income, calculating initial weights of three targets of economy, low carbon and reliability, and carrying out optimized search by taking the initial weights as the center to generate dynamic weight vectors; Acquiring updated operation data, constructing a system state space and a scheduling action space based on the dynamic weight vector and the updated operation data, defining a multi-objective instant cost function and an approximate cost function, and forming an approximate dynamic planning scheduling model; Based on the approximate dynamic programming scheduling model, performing daily scheduling optimization and daily rolling scheduling optimization, outputting scheduling instructions, and updating parameters of the approximate cost function in the daily rolling scheduling optimization process; And after the scheduling instruction is executed, acquiring the