In order to improve the ability of the Active Power Filter (APF) to suppress the harmonics and compensate the power quality in the operation of the micro-grid, a recursive integral PI-repetitive control strategy for BP neural network for APF in micro-grid is proposed. Combined with the parallel APF topology, BP neural network and recursive integral function are introduced on the basis of traditional PI control. According to the trend of tracking error change, the PI control parameters are adjusted in real time and quickly, and then the optimization requirements are met. In parallel with the repetitive control, the ability of tracking steady state error is improved to ensure the stability of the system operation. The parallel APF simulation model and experimental device are established. Simulation results show that the proposed control strategy can improve the response speed and the compensation precision, improving the robustness of active power filter. This work is supported by Foundation of Liaoning Education Department (No. LJYL016).