Multi-agent Collaborative Optimization Method Of Virtual Power Plant Based On Bayesian Game

The continuous increase in the proportion of Distributed Energy Resources(DERs)poses challenges to the power system.In response to this challenge,the concept of virtual power plants(VPP)comes into being.It relies on advanced communication software and demand response technology and has become an effective energy management method of DERs.The virtual power plant plays an important role in the dispatching operation of the power grid.Considering various types of DERs in the virtual power plant may belong to different property rights entities,game theory is usually used to describe the multi-agent interaction optimization process when virtual power plant participates in the dispatching of power grid.Most of the existing researches on optimization modeling are based on complete information.However,in the actual scheduling operation of virtual power plants,due to privacy protection and some other reasons,there are often scenarios where some game players are unwilling to share private information or some information cannot be obtained.At this time,the complete information game theory will no longer applicable.Therefore,it is necessary to study the incomplete information game theory to analyze the incomplete information optimization problems in the virtual power plant scheduling operation process,so as to realize the multi-agent collaborative optimization of VPP.In this paper,a multi-agent collaborative optimization method of virtual power plants based on Bayesian game is proposed.Firstly,considering VPP participates in grid dispatching by technical virtual power plants,the multi-agent interaction optimization framework for VPP is introduced from the perspectives of external dispatch and internal dispatch respectively,and the basic models of DERs aggregated by VPP,such as controllable and uncontrollable distributed generation,energy storage systems and demand response load,are established under this framework.Secondly,the Bayesian game theory and its main analysis method the Harsanyi transformation are introduced.On this basis,focusing on the internal dispatching of the VPP and taking the demand response load as the analysis object,the scenario where virtual power plant operated by the power grid aggregate intelligent communities to conduct the day-ahead energy optimization dispatching is formulated based on incomplete information,and profit models of VPP operators and residential intelligent communities are established respectively.Then,through using Bayesian game theory,a virtual power plant multi-agent collaborative optimization model is established to describe the interaction optimization relationship between intelligent communities in the virtual power plant of this scenario,and the existence and uniqueness of Bayesian Nash equilibrium is proved.Finally,an iteration algorithm for solving Bayesian Nash equilibrium is designed,and the case study of an virtual power plants and simulation results is provided to verify the effectiveness of the optimization model and algorithm.