For solving the problem that the large rotating machinery was disturbed by the rotor imbalance and external complex environment excitations during the operation，which caused the rotor system to vibrate，and generated multi-frequency transmission forces to the base and external machinery，an adaptive PD control based on BP neural network algorithm was proposed. A hybrid bearing structure integrating electromagnetic actuator and journal bearing was applied，and its dynamic characteristics were analyzed. The finite element method was used to establish a multi-span rotor system dynamic equation，and the transmission forces control under PD control was analyzed in principle. For the defects of traditional PID control，an adaptive PD control algorithm based on BP neural network was proposed. Numerical simulations were carried out in a four-bearing double-span rotor system and effectiveness of adaptive PD control based on BP neural network，BP neural network control and LMS control were compared respectively. The results showed that the adaptive PD control based on BP neural network has better suppression effect on the multifrequency transmission forces control of the rotor system.