Model-data-event based community integrated energy system low-carbon economic scheduling

With environmental pollution becoming an increasingly serious problem, it is crucial to develop renewable energy and improve energy utilization efficiency. Integrated energy system realizes complementary utilization of various energies by integrating various heterogeneous energies. However, the uncertainty of generation and load causes random fluctuation, making it difficult to accurately and quickly adapt to the dynamic environment. Therefore, this research designs a model-data-event based low-carbon economic scheduling framework for the community integrated energy system. Firstly, the low-carbon economic scheduling problem is mathematically described, then the decision-making problem is expressed as a reinforcement learning framework. An improved DDPG algorithm which can cope with the uncertainty is used to make real-time scheduling decisions. Considering the instability of reinforcement learning algorithm, this research proposes a heuristic real-time power allocation strategy, which comprehensively considers the degree of imbalance and the schedulable ability of the responders. The event-driven mechanism enables the responders to respond quickly and accurately to the energy compensation requirements. Simulation results indicate that the proposed scheduling framework can adapt to different parameters and reduce energy costs up to 6.65% and 3.85% compared to the deep Q network and soft actor critic algorithms, with the computational efficiency 11 times higher than that of model predictive control algorithm.