Collaborative Optimization Model for Multi-Park Integrated Energy System Based on Dynamic Carbon Trading Mechanism

To address the challenges of multi - agent interest conflicts and low - carbon collaboration in the multi - park integrated energy system (MPIES), this paper proposes a two-layer hybrid game optimization model that integrates dynamic carbon trading and deep reinforcement learning. First, a dynamic carbon price adjustment mechanism and a green certificate-carbon emission elastic offset model are constructed to strengthen the guiding role of market supply and demand in carbon trading. Second, a two - layer game architecture of “Independent System Operator (ISO)- park” is designed. The upper layer uses the Deep Q - Network (DQN) to dynamically optimize the energy pricing strategy, and the lower layer realizes the synergy between the park equipment output and demand response through the Proximal Policy Optimization (PPO) algorithm. Finally, a Shapley value benefit distribution method weighted by carbon contribution is proposed. The simulation results show that the model can reduce the system’s carbon emissions by 24.8%, increase the alliance’s economic benefits by 18.7%, and the convergence speed of the algorithm is 35% faster than that of traditional methods. This research provides theoretical support for the low - carbon collaborative operation of park clusters under the “dual carbon” goal.