Abstract: The large-scale construction of microgrids with large-scale distributed power sources and the application of the vehicle-togrid（V2G） technology lead to the instability of the operation voltage of the distribution network and the demand loss of electric vehicle（EV） users, and also bring a new means for the regulation of the power system. Therefore, this paper proposes a bi-layer active power-reactive power coordinated control strategy based on enhanced evolutionary-deep reinforcement learning（EDRL） for distribution networks and microgrids with V2G. Firstly, considering the influence of V2G process on the demand of EV users, a bilayer coordinated control model including distribution networks and microgrids with V2G is constructed based on the travel chain.Secondly, the enhanced EDRL algorithm is constructed to further enhance the convergence ability of the agent. Then, defining the operation information of the distribution network as the state set, the power regulation signal of each unit as the action set, and the comprehensive cost such as voltage deviation, network loss and user demand loss as the reward function index, the structure design of the bi-layer coordinated control is completed. The case results show that, the proposed strategy can reduce the voltage deviation and network loss of the distribution network on the premise of meeting the charging demand of EV users.