Some numerical experiments with nonlinear model predictive control

摘要

This paper focuses on implementation aspects of nonlinear model predictive control (NMPC) and, based on three applications of increasing complexity, addresses a number of practical issues. The influence of the control signal parametrization is illustrated with the first application, in which NMPC is applied to achieve maximum conversion in a continuous stirred tank reactor. In the second application, an observer-based predictive control scheme is designed for a polymerization reactor. This example allows the effect of modeling errors and their compensation to be discussed. Finally, control of nonlinear distributed parameter systems is considered in the third application.