Gain-scheduled state estimation and optimizing nmpc applied to a non-minimum phase CSTR

In model predictive control techniques, the reference values are usually adjusted by an infrequent optimization of a rigorous nonlinear stationary plant model. Between these optimizations however the process may undergo sub-optimality of operation due to the presence of the disturbances. Lately, nonlinear model predictive control based on online dynamic optimization has been proposed to combine optimal operation and feedback control. In this work, several different NMPC control strategies are considered. These controllers are coupled with a nonlinear gain-scheduled state estimator to provide output feedback. The potential of the optimizing as well as of an economics-oriented tracking control schemes are demonstrated for a non-minimum phase continuous stirred tank reactor (CSTR). The output of the controller in both cases is an economics-driven response since the profit function is considered in the objective criterion. A comparison of the performance among the conventional tracking, the economics-oriented, and the optimizing control is shown.