Feedback action in predictive control: An experimental case study

A state-space predictive controller is applied for control of a gas conditioning unit, an integrating process with two inputs and two outputs, in a real-life environment. The controller is used for low-level control where feedback performance is very important. Disturbance model structure selection is discussed in detail. The tuning procedure for the Kalman filter, based on the sensitivity functions approach, is explained through the study of simplified limit cases with few tuning parameters. Robustness is verified with a set of possible true models. While the conventional model predictive control approach results in much slower feedback performance than a simple diagonal PI control scheme, the proposed controller rejects load disturbances more efficiently.