Abstract: In order to realize the on-line adjustment of PID controller parameters, the BP neural network PID controller is designed to adjust the PID parameters in real time, by using the adaptive self-learning ability of the BP neural network. Based on the state space model of the research reactor core and the design principle of the reactor core control system, the BP neural network PID control system of the research reactor core power is established, and the core dynamic simulation is carried out. The results show that compared with the traditional PID controller, the BP neural network PID controller has the characteristics of small overshoot, fast response and better control effect, which is suitable for research reactor core power control.