Because the robot weapon station servo system has nonlinear and non-determinacy character such as clearance and friction, combine the PID controller and the fuzzy-neural network. Using the neural network to get hold of fuzzy rules and construct the fuzzy neural network controller. It can adjust the three parameters of PID control online according to error and error variance. Simulation results showed that this controller strategy has the advantage of PID controller with high accuracy and the characters of fuzzy neural network PID controller with high speed, minor overshoot, good stability and better dynamic and static performances.