With the deepening of China's electricity market reform,competition in the power generation market is becoming increasingly fierce,and power generators are facing multiple uncertainties and challenges.To this end,a clearing model for the electricity spot market was first constructed as the external environment of the multi-agent model.Then,the deep deterministic policy gradient(DDPG)was placed under the multi-agent(MA)system for natural extension to obtain the MADDPG model.Finally,this model was used to improve the training effectiveness and strategy optimization ability of the agents in the multi-agent environment,enabling power suppliers to maximize profits in competition through learning and optimizing strategies.The results show that the MADDPG bidding strategy model calculates electricity prices in the A and B generation node systems,and tends to stabilize after 50 and 48 iterations,respectively.In addition,the MADDPG bidding strategy model can accurately predict the load of the electricity market,with a maximum error of no more than 2%between the predicted value and the actual value.The MADDPG bidding strategy model demonstrates good stability and predictive accuracy in optimizing electricity market bidding,effectively helping power suppliers maximize profits in fierce market competition.
维普
万方数据