A modified Smith predictor with a new structure for unstable processes

Abstract

In this paper, we propose a modified Smith predictor with a new structure for unstable processes. The proposed new structure makes it possible to use the unstable model itself as a process model, while the previous modified Smith predictors provided no option to use a mismatched stable process model to guarantee stability. Accordingly, the proposed Smith predictor can predict the dynamics of the actual process much better than previous approaches. It also guarantees good robustness to modeling errors by separating servo and regulatory problems. Moreover, it demonstrates good disturbance rejection performances because the adjustable parameters of the PID controller for the regulatory problem are tuned systematically with an internal feedback loop.