Electricity-heat collaborative optimization strategy in microgrid using deep reinforcement learning

With the massive penetration of renewable energy, the flexibility of microgrids is rapidly declining, and its optimal operation is facing great challenges. For the microgrid that uses regional centralized heating as the source of heating power, we propose to use local electric heating devices to provide auxiliary heating to reduce the operating cost of the microgrid. We first establish an electricity-heat collaborative optimization framework that considers real-time prices in the electricity market and unit heating power prices in regional centralized heating. Then, in order to minimize the long-term cost of the microgrid, we transformed the optimized operation of the microgrid into a Markov decision process problem, and applied the deep deterministic policy gradient algorithm to solve the problem. Finally, we verify through simulation experiments that the architecture and algorithm proposed in this paper can effectively reduce the operating cost of the microgrid by 27.5%, and the algorithm has good convergence and stability.