Nash Q-learning based equilibrium transfer for integrated energy management game with We-Energy

This paper proposes an innovative energy interacting unit (We-Energy) with the characteristic of full duplex trading mode. In order to manage all the We-Energies in an optimal way, a new integrated energy management framework based on a noncooperative game is performed so as to allocate the energy demands of each WE such that the benefit of each WE can be maximized. To overcome the impact of the randomness and inaccurate information of renewable energy sources, Nash Q-learning algorithm is applied for computation of game equilibrium under the unknown environment. The novelty of the proposed algorithms is related to the incorporation of the continuous action space into the discrete adaptive action set and combined the equilibrium transfer to improve the efficiency of the algorithm. Simulation studies of modified IMS confirm that it has a better performance with the desired equilibrium strategy and convergence speed. (C) 2019 Elsevier B.V. All rights reserved.