Research On Bidding Strategies Of Generation And Demand Side Based On Deep Q Learning In The Energy Market

In recent years,with the implementation of electricity market reform,China has accelerated the process of pilot power market design and operation.At present,a number of pilot provinces have practically started the trial operation of the spot market,such as Zhejiang,Guangdong etc.In the design of market rules in various regions,it is proposed to adopt twoside bidding,namely demand-side bidding included that meets with the law of market resource allocation in the future.Compared with the power market operation mode of one-side bidding,namely only generator companies bidding included,the two-side bidding mode is more preferred and efficient,which the complexity of the market operation system and market rule design is also obvious.Under the two-side bidding mechanism,how will power generators change their bidding strategies,how will power purchasers with the power generation group background perform in the market,and what new problems will arise when market operators are concerned about the use of market power.In response to the above problems,the transaction behavior of market entities and the overall operation of the market under the two-side bidding mode are simulated and analyzed.The bidding model of power generators and purchasers is proposed and the strategic behaviors of these two types of market entities in different scenarios are simulated.Taking into account the possible emergence of power groups integrating power generation and sale after the reform of Chinese electricity market,the collusion behavior of market entities is simulated.The specific work is as follows:First,the bidding model of generators and electricity purchasers under the two-side bidding mode in the spot market and the market clearing model are established,and the operation index evaluation system of the generation side and the market side is considered.The market clearing results under the two bidding modes and the impact of two-side bidding on power generation and market operation are analyzed.Then,on the basis of the above bidding model,considering the market environment of unknown information,with the goal of maximizing own profits,a strategic bidding model of market entities based on the deep double-Q network algorithm is proposed to simulate individual strategic behavior,the bidding strategies of generators and electricity purchasers and the overall market operation under different bidding rules and network parameters are analyzed.Finally,considering the policy background of the power generation group to establish power groups integrating power generation and sale in the future,the concept of slight altruism is introduced.Based on the above bidding model that does not consider collusion,a collusion bidding model for market members in the power groups integrating power generation and sale is established,and the bidding strategies of the members of the power groups integrating power generation and sale under different collusion rules are simulated.