Feedback linearization is a nonlinear controller design strategy that results in an explicit formulation of the feedback control law. This method can result in excellent performance if an accurate dynamic process model is available. However, feedback linearization suffers from a lack of robustness if plant-model mismatch exists. The approach presented in this work analyzes the robustness properties of the closed-loop process with specific regard to the controller tuning parameter. Due to this, it is possible to tune the controller such that robustness over the entire operating region is guaranteed even under the assumption of certain types of model mismatch. This method is illustrated with an example and conclusions about its applicability to more general model and controller formulations are presented.
Accepted to Proceedings of the ADCHEM 2003, Hong-Kong, China (2003)