Decoupling neuro-fuzzy model predictive controllers applied to quadruple tanks

摘要

This paper is focused on the design of an algorithm for decoupling multivariable systems based on generalized predictive control and its application to control of a quadruple tank system. The quadruple-tank process is representative of a multivariable control problem that may be found in a chemical engineering process control application. To handle the nonlinearities, a Takagi-Sugeno neuro-fuzzy system is suggested to model the plant dynamics at relevant operating regions. The developed model is used as a predictive model for a decoupling model predictive control algorithm.