Lyapunov-based MPC for nonlinear process with online triggered linearised model

Most of industrial processes are nonlinear. Model predictive control (MPC) using an explicit nonlinear model can achieve satisfactory performance, however, it will bring a high computational burden. Although linear MPC is widely used in practice, a linear model cannot deal with the highly nonlinear system dynamic that is well overall in a wide operating region. In this study, an error trigger rule evoking a re-modelling algorithm to re-linearise the known nonlinear analytical model has been proposed for closed-loop nonlinear systems with input constraints. The error-triggering can be conducted by an error quantiser that quantifies model error and the re-linearisation program is triggered when the accumulated error exceeds the set threshold. The stability of the process is maintained by using the Lyapunov-based MPC. The effectiveness of the proposed control algorithm is validated by a chemical process simulation.