A Simple Controller via Feedback Linearization and its Use in Adaptive Control

In this paper it is shown that if an input-output discrete-time model is available, feedback linearization can be easily accomplished without the use of geometric differential tools and without the need for measurements of all states or state observers. Feedback linearization is sub-sequently combined with internal model controller design to obtain a nonlinear controller that can stabilize unstable steady states, follow setpoint changes and reject disturbances. Global stability is proved in the absence of plant/model mismatch and zero steady-state offset is guaranteed if the closed loop system is stable, even for inaccurate models. Subsequently, the above controller is used to develop a nonlinear adaptive controller. The effectiveness of the adaptive controller is demonstrated by application to a highly nonlinear simulated continuous stirred-tank reactor.