Aiming at the control problem of electromagnetic bearing system, in order to improve the dynamic characteristics and anti-interference ability of the system, a single degree of freedom mathematical model is established based on electromagnetic bearings, and an adaptive PID controller is designed based on an improved RBF neural network and Adam algorithm to adjust the system hyperparameters in real time. The experimental results show that under the given parameters of electromagnetic bearings, the adaptive PID controller reduces the maximum overshoot of step response by 26.68% compared to conventional PID controllers; After being subjected to the same 0.1 mm disturbance, the maximum offset of the system decreased by 21.78%, verifying that the controller has good control effect and robustness. The control effect of the electromagnetic bearing system is superior to the conventional PID control algorithm.