Research On Optimal Scheduling Of Multi-agent Integrated Energy System Based On Game Theory

In order to achieve the "double carbon" goal,increasing the proportion of renewable energy power generation is the only way,which will lead to problems such as increased uncertainty in power system output.The new-generation power system represented by the Internet will become the protagonist of the future power grid.This topic is driven by a photovoltaic power station dataset in Australia,and takes the coordinated operation of multi-integrated energy systems as the core of the research.It proposes different types of operation frameworks and conducts operational economic analysis and equipment output analysis on them.The main research content and results as follows:First,the power output of photovoltaic power plants is predicted based on the Cat Boost algorithm.Use the photovoltaic output data set to predict the photovoltaic output in the day ahead,and compare it with BP neural network,decision tree and other algorithms.The results show that the evaluation index of the proposed algorithm,such as nRMSE,increases by an average of 1.346% in the comparison of sunny standard days,and an average increase of 0.733% in the comparison of different weather types.At the same time,the prediction speed and generalization ability can meet the engineering needs.Secondly,a multi-integrated energy system coordination and optimization operation model based on master-slave game is proposed.First,the model is divided into upper-level energy managers and lower-level integrated energy system users.The upper-level sells electricity and optimizes electricity prices,and the lower-level purchases electricity and feeds back energy use plans.Finally,it iterates until both parties reach a balance of interests.The results of the calculation example show that after the optimization of the model,the operating cost of the lower-level users is reduced by 9.41%,and the income of the upper-level energy managers also increases by171.83%.Finally,a cooperative game-based peer-to-peer transaction model for multiintegrated energy systems is proposed.First of all,considering the situation that multiple integrated energy systems need to interact with each other without negotiating prices with energy managers,a framework of point-to-point trading energy systems combining cloud and edge is constructed.Based on this framework,a multi-integrated energy system alliance is proposed Nash bargaining model,and use the cooperative game theory to transform the model into the sub-problem of minimizing the operation cost of multi-system alliance and the sub-problem of income distribution,which are solved by the method of alternating direction multipliers.The results of the calculation example show that after calculating the income distribution sub-problems,the seller’s operating cost is reduced by 5.311%,on the contrary,the buyer’s cost is increased by15.14%,and the rationality of income distribution is further increased.At the same time,the total operating cost of each integrated energy system is lower than that of Playing cooperative games decreased by 2.34%.To sum up,the research on the optimal operation of multi-integrated energy systems based on game theory in this paper provides ideas for the organization and effective operation of multi-systems,which has certain theoretical value and engineering significance.It provides a reference for business ideas and can also provide some reference for policy formulation.