Non-Cooperative Bidding Strategy in Electricity Markets Based on Multi-Agent Deep Deterministic Policy Gradient

With the increasing complexity of electricity markets, traditional bidding mechanisms are facing numerous challenges, particularly under the influence of diversified energy structures and competing market participants. This paper proposes an optimization approach for electricity market bidding strategies based on Muti-agents Deep Deterministic Policy Gradient (MADDPG). By modeling multiple types of market participants, including thermal power units, renewable energy units, and energy storage units, we employ a deep DDPG network to simulate the bidding decision-making process of these participants in the electricity market. Experimental results demonstrate that the reinforcement learning-based bidding strategy can effectively enhance the profitability of market participants while exhibiting strong adaptability in dynamic market environments. This study provides a novel solution for intelligent trading and risk management in electricity markets.