Wet flue gas desulfurization system is a desulfurization process which reduces flue gas SO 2 content by acid-base reaction and PH value is the main control parameter. Since the system possesses time-varying, large inertia and nonlinear characteristics, the traditional PID algorithm is difficult to obtain satisfactory results in condition that a high demand of steady-state performance and control accuracy. So this paper proposed the method of RBF neural network online tuning PID control. The PID control is simple and easy to be used. The RBF network can tune PID parameters in real time based on the change of object parameters. The method combines both methods. The simulation results show that RBF neural network PID controller is better than the traditional PID controller, in the control effect of the system. It has a small amount of overshoot, short adjusting time and strong robustness, etc.