Real time energy management and control strategy for CCHP microgrid based on ADHDP

Considering the fact that the optimization of operation process of CCHP microgrid is complex, characterized with uncertainties, nonlinearity, strong coupling and dynamics. This paper proposes a method for optimal operation of CCHP microgrid system based on ADHDP (action based heuristic dynamic programming) to solve this problem. ADHDP involves three neural networks, namely the Action Network, Critic Network and Model Network. The Critic network weights keep being updated inline with the control error and performance index function. In addition, the action network weights keep being updated inline with the control error. The ADHDP model is trained with real-time data of the system until an optimal control strategy of CCHP microgrid is obtained. In the end, the simulation results suggest that the proposed scheduling method has effectively reduced the total electricity cost and improved energy efficiency.