Multiple Model Predictive Control Based on Fuzzy Switching Scheme of a Coagulation Chemical Dosing Unit for Water Treatment Plants∗

A switching scheme based on Takagi-Sugeno fuzzy inference system is proposed in this paper to address the problems of poor transient response, rapid oscillation of plant output and long duration of switching time associated with the multiple model predictive control (MMPC) based on hard switching. A set of piecewise linear models is used to represent the system under consideration at different operating regimes. Corresponding MPC local controller is developed for each model. At each instant, the fuzzy switching system selects the appropriate model/controller pair for the system. The proposed MMPC strategy is applied to a coagulation chemical dosing unit for water purification plants. Simulation results of the proposed control scheme are promising and positive.