Dynamic characteristics of the mirco/nano-transmission system are analyzed. After simplified micro-nano-positioning platform to be a spring-mass-damper model equivalently，its dynamic model is established using Lagrange approach. Based on the capacitor characteristics of piezoelectric actuator (PZT)，the overall-transfer-function of the system is deduced. Because of self-training function，adaptability and nonlinear approximation of BP neural network，the PID parameters are self-tuned online. A BPNN-PID controller is designed by combining the advantages of simpleness，easy realization and good robustness of PID method. The BPNN-PID controller realizes higher control accuracy and shorter steady-state residence time compared with traditional PID，and well meets the need of high accuracy and fast response of the mirco/nano-transmission system.