本文探討了能源市場中的拍賣行為，使用多代理模型進行模擬。我們將電力供應商和消費者建模為自主代理，他們在多代理環境中做出決策以最大化其效用。然而，由於代理之間的信息不足，每個代理都難以實現其最佳決策。為了解決這個問題，我們提出使用納許Q學習，它結合了納許均衡和Q學習，以在考慮其他代理出價行為的同時最大化每個參與者的效用。在多個案例研究中，我們證明了納許Q學習算法能夠確保參與者最終達到納許均衡。;This paper explores auction behavior in the energy market using a multi-agent model. We model electricity suppliers and consumers as autonomous agents who make decisions to maximize their utilities in a multi-agent environment. However, due to insufficient information between the agents, each agent faces difficulty achieving his/her optimal decision. To address this issue, we propose using Nash Q-learning, consisting of Nash equilibrium and Q-learning, to maximize each participant′s utility while considering the bidding behavior of the other agents. In several case studies, we demonstrate that the Nash Q-learning algorithm ensures participants eventually reach the Nash equilibriums.
