The traditional sweeper swing arm open-loop control system cannot adapt to high speed cleaning operation,because of its insufficient dynamic response characteristics and weak anti-interference ability.In order to solve problems of local extreme value and slow convergence of BP neural network(BPNN),an improved BPNN PID algorithm was proposed,the core of which was to suppress influence of previous PID output u ( k -1) on output u ( k ) through active series correction.A high-speed sweeper swing arm actuator was introduced,a closed-loop control system using BPNN PID controller was designed.By building Simulink-AMESim co-simulation model,the step response,anti-interference ability and position tracking ability of control system were studied.The improved BPNN PID controller can adjust PID parameters dynamically,and improve adaptability,accuracy and stability of system.The improve BPNN PID controller has stronger anti-interference ability and better robustness.In addition,the system has almost no overshoot,the overshoot is 0.5%,which is only 2.34% of PID controller,the stability time is 0.62 s,which is 66.31% ahead of PID.