Multilateral Decision Strategy Of Electric Vehicle Charging Market Based On Multi-Agent Game

The rapid growth in the number of electric vehicles（EVs）poses difficulties and challenges for the electric vehicle charging market.Besides,the additional demand brought by EVs when they are connected to the grid can cause a significant impact.Under the existing power supply strategy,the power supply in the EV charging market does not match the power demand of EV users,which may lead to loss of profit and total social welfare of all parties in the charging market.How to design a strategy to maximize the profit of all parties involved in the EV charging market and to maximize the total welfare of the society is an important part of the smart grid research.Traditional smart grid optimization schemes often consider only the unilateral model of supply and demand side but ignore the other participants in the market,resulting in compromised optimization results.Based on this,this thesis constructs a three-party EV charging market model based on historical data and solves the optimal strategy for each party under the specified objective to maximize the benefits of each party.In this thesis,we first build a three-party charging market model,which consists of the following main participants: charging vehicle users,aggregators and the grid.Each participant makes strategy based on its own revenue and feeds back to the market based on its observation of the market.In the market,EV charging customers observe the price of electricity offered by the collector and choose their own demand based on their own utility maximization principle.The grid observes the demand curve of customers in the market and sets the supply quantity and wholesale tariff based on its own revenue and operating cost.The collector observes all the information in the market and sets the retail price based on its own operational efficiency.Second,this thesis proposes two hypothetical objectives to solve the optimal strategy for each party involved in the market,and compares the solution results in terms of single-party benefits,social welfare,and market operation conditions.Among them,hypothesis one aims at the decision of each participant under the objective of maximizing the total profit of a single party,and hypothesis two aims at the decision of each participant under the objective of maximizing social welfare.Traditional algorithms for solving multi-objective optimization are fast but difficult to cope with immediate factor changes.For the multi-party decision problem in charging market,deep reinforcement learning and multi-intelligent games can solve this problem.At the same time,multi-intelligent reinforcement learning can effectively deal with the multi-party game situation.welfare.This work will be validated under a virtual environment,the benefit and cost of each party under different assumptions will be quantified.