Input-Output linearization using approximate process models

An approach to approximate feedback linearization is presented which utilizes an approximate input-output normal form model for the process. The approach is valid for minimjm phase systems, and yields a dynamic output feedback controller of order 2N − R + 1, where N is the dynamic approximation order and R is the relative degree of the approximation model. Conditions are presented under which local (linear) stability can be guaranteed despite the plant-model mismatch. Two reactor control case studies are presented to show how the approach can be used both when a fundamental process model is available and when only process input-output data are available.