Abstract: The exploration and utilization of the reactive power response potential of electric vehicle charging piles can optimize the operation of the distribution network and increase the utilization rate of infrastructure capacity. In this paper, a distributed model predictive control strategy in which residential charging piles participate in reactive power response is proposed. On the premise of not relying on control and communication centers, peer-to-peer（P2 P） communication and edge computing are used to realize the coordinated reactive power response of charging piles and improve the voltage quality of the distribution network. First, based on the instantaneous power theory and the sensitivity matrix, an optimal model of the reactive power response of the charging pile is established. Model predictive control algorithm is used to solve the optimal setting value of control voltage for the charging pile.Then, a neighborhood P2 P communication strategy is established based on the asynchronous replacement of timestamps. The voltage and power prediction information of each other is exchanged to realize the decoupling of optimization problems for multiple piles and improve the overall voltage regulation effect. Finally, an example simulation shows that the proposed reactive power control strategy can effectively reduce the grid voltage deviation on the premise of meeting the active charging demand of the charging pile. In addition, in the P2 P communication interaction mode, the voltage regulation effect of the proposed strategy is significantly improved compared to the disorder response of the charging pile in the non-communication mode, and it has better communication delay robustness.