Bidding Strategy Optimization for Generation Companies Based on Multi-agent Reinforcement Learning Considering the Impact of Long-term Market

This paper introduces a method for optimizing bidding strategies of generation companies in long-term market and spot market, based on multi-agent reinforcement learning. This paper first proposes a mathematical model for generation companies' decision-making in long-term trading and spot market bidding, then proposes an optimization method for units' bidding strategies, and finally solves the bidding strategies optimization problem using a multi-agent reinforcement learning approach. A case study based on actual data of province A in China is provided to illustrate the relevant method. This paper optimizes the bidding strategies of various types of power units in different long-term price scenarios. The results indicate that higher long-term prices lead to a higher proportion of long-term electricity volume. This study also analyzes the profitability of generation units in the long-term market and the spot market under different scenarios.