The traditional PID neural network (PIDNN) limited the scope of application and integration problems are difficult to obtain the error rule. For the realization of nonlinear multivariable control systems, neural network control system to expand the application range of this paper, based on improved version particle swarm optimization algorithm for PID neural network control system design solution, replacing the traditional BP back the propagation algorithm, simulation results show that compared with traditional PIDNN, the steady-state system, robustness and accuracy have improved obviously, this method is effective to improve the use of PID control, intelligent method for the PID Control proposed a new reference.