Resource Allocation in Non-Cooperative Game Models with Reinforcement Learning

This study proposes a non-cooperative game model that includes a two-layer supply and demand relationship and employs reinforcement learning for its solution. The study takes the water resource allocation in Pujiang County, Jinhua City, Zhejiang Province, as a case study, constructing a non-cooperative game model with reservoirs, water plants, and fac-tories as the main entities. It simulates the strategic interactions among various water users to optimize the water resource allocation plan. This model takes into account the scarcity of water resources, the dynamic changes in demand, and the fairness of distribution. By introducing the CommFormer reinforcement learning algorithm, the model achieves policy optimization in complex and dynamic environments, reduces computational complexity, and speeds up decision-making response time.