No abstract

No abstract

A strategy based on Nonlinear Programming (NLP) sensitivity is developed to establish stability bounds on the plant/model mismatch for a class of optimization-based Model Predictive Control (MPC) algorithms. By extending well-known nominal stability properties for these controllers, we derive a sucient condition for robust stability of these controllers. This condition can also be used to assess the extent of model mismatch that can be tolerated to guarantee robust stability. In this derivation we deal with MPC controllers with ®nal time constraints or in®nite time horizons. Also for this initial study we concentrate only on discrete time systems and unconstrained state feedback control laws with all of the states measured. To illustrate this approach we give two examples: a linear ®rst-order dynamic system and a nonlinear SISO system involving a ®rst order reaction. #

Recently, the unscented Kalman filter (UKF) algorithm, which is a new generalization of the Kalman filter for nonlinear systems, was proposed in the literature. It has significant advantages over its widely used predecessor, the extended Kalman filter (EKF). These include better accuracy and simpler implementation and the dispensability of system and measurement model differentiability. In this work, we compare the performance of the two approaches in a simulated pH process with three situations considered. The first one evaluates the performance differences between the unscented transform and the EKF linearization, as applied to the nonlinear pH output equation. In the second simulation, the complete filter algorithms are compared in a state estimation framework. The third case concerns parameter estimation. In all three cases, the UKF produced more-accurate results.

No abstract

In this work a robust nonlinear model predictive controller for nonlinear convection-diffusion-reaction systems is presented. The controller makes use of a collection of reduced order approximations of the plant (models) re-

No abstract

A nonlinear model predictive control algorithm is implemented on-line to control the liquid level and temperature in a pilot plant CSTR, where an irreversible exothermic chemical reaction is simulated experimentally by steam injection. The dynamic behavior of the pilot plant reactor is represented using a mechanistic, first principle model and a comparison between off-line simulation and experimental data is presented. Several sources of model mismatch and unmeasured disturbances are identified that affect the quality of the model in representing the reactor dynamics. Despite these mismatches and disturbances, the closed loop system is able to track setpoint changes and reject disturbances quite well. In particular, the NMPC controller is demonstrated for different tuning parameters and under conditions of constraint saturation at unstable points. r