A simple modified Smith predictor controller design was proposed for integrating processes with long dead time. The overall control structure introduced two physically meaningful tuning parameters. One was used to set the speed of the closed-loop servo response and the other was used to set the performance of the load response. A simple controller tuning method was also developed to obtain the PI tuning parameters. With a given estimation of the unmatched model, the stability region of the tuning parameter was estimated. Two examples were used to demonstrate the effectiveness of the method.

A predictive control method based on differential evolution was put forward in connection with changing parameters nonlinear model. By using neural network and accord as input and output data of the system, a predictive model was established. And then differential evolution algorithm was used to do rolling optimization and improve system performance accordingly. At last, GA and DE were used respectively to do predictive simulation to realize the rolling optimization and compare the two results. The two simulation results show that the rolling optimization of predictive control based on DE is better than that based on GA on the whole.