OVERVIEW

In the approach for feedback control outlined earlier in this book, very little was required in the way of process knowledge. The only knowledge that was absolutely necessary was the directional response of the process to a change in controller output so the direct or reverse action of the controller could be properly set. Other than that, controller tuning, however determined, was used in feedback control to match the controller to the process. Indeed, the remarkable feature of the PID controller is its ability to bring the measurement to a desired value in the presence of set point or load changes with only minimal knowledge of the process. This feature is the contribution of the integral mode.

However, if knowledge of the process, and specifically some type of process model, is taken into account more precise controllers can be designed. Simple model-based controllers have been in use for some time. Currently, the amount of activity in developing and applying more sophisticated forms of model-based control is increasing. The motivation behind it is the desire to provide better control for more difficult applications, such as processes with a long dead time relative to the dominant time constant, processes with unusual response characteristics such as inverse response, processes in which control loops interact significantly, and processes that are subject to multiple physical or operational constraints.

This chapter presents an overview of model-based control. We start with the Smith predictor controller, which has been in use for many years, and proceed through...