Abstract:

Multi-microgrid(MMG) system can provide reliable and independent power for load center. However, MMG system is faced with problems such as real-time management, economical operation and control. Based on this, a new energy management system(EMS) is proposed, which transforms MMG of multiple different stakeholders into a unified and efficient system. In order to ensure that each MMG can achieve its own operation objectives, the EMS proposed in this paper is based on cooperative game to achieve efficient and coordinated operation of MMG system and ensure fair distribution of energy costs among alliance members. In addition, the energy cost distribution problem when the number of alliance members increases exponentially is fully considered, which uses column generation algorithm to solve the above problems. In addition, deep convolutional neural network(CNN) is used to estimate the daily operating cost of MMG, and a scheduling strategy is proposed to optimize the daily total operating cost of MMG. Finally, in order to verify the effectiveness and superiority of the proposed model, it is compared with the existing optimal scheduling strategies, such as approximate dynamic programming(ADP), model prediction control(MPC) and greedy algorithm. The simulation results show that each MMG can save energy and reduce consumption through cooperative game, and the daily operating cost of the proposed optimization strategy is significantly lower than that of other methods.

Key words: Column generation algorithm, cooperative game, convolutional neural network, energy management system, multi-microgrid, renewable energy sources