Evaluation of the NEPSAC nonlinear predictive controller on a thermal process

Abstract

Nonlinear dynamics are commonly encountered in industrial applications, where manufacturing of higher quality products very often requires that the process works within a wide range of operating conditions close to the boundaries. Nonlinear Model Predictive Control (NMPC) appears as a solution due to its capability to find optimal control actions for the case of nonlinear processes with constraints. In this contribution, the control problem is solved using the Nonlinear Extended Prediction Self-Adaptive Control (NEPSAC) approach to model predictive control (MPC), which besides of being a fast algorithm also avoids explicit local linearization by directly using the nonlinear model for prediction. The effectiveness of the mentioned nonlinear controller and the procedure to express a nonlinear model suitable for prediction is illustrated on a simulation example of a highly nonlinear thermal process. Furthermore, the benefits of NEPSAC are clearly shown by comparing its performance to linear controllers such as linear MPC, PI and PID controllers.