Design A Nonlinear Model Predictive Control for Coupled-tanks Systems

Model predictive control is a control approach that is commonly used in industrial processes. However since the majority of predictive controllers have been designed based on linear models of systems, their control performances are limited when systems are working over wide operating regions. This paper presents a design study of predictive controllers based on nonlinear models of systems. The proposed nonlinear models are Takagi-Sugeno fuzzy models obtained by means of system identification based on input-output measurement data. The Levenberg-Marquardt nonlinear optimization algorithm is suggested to compute optimal control signals in the predictive controller. The design study was applied to a coupled-tanks system. Simulation and real time control results show the proposed nonlinear predictive controller is able to deal with control performace requirements better than that of other linear and nonlinear predictive control method