Non-linear Model Predictive Control of a Quadruple-Tank Process

This paper presents a non-linear model predictive controller for offset-free tracking and disturbance rejection of arbitrary constant (or piecewise-constant) set-points and/or disturbances. The control problem consists of regulating the non-linear plant dynamics augmented with the integral of the tracking error of the variables to be controlled. This simple approach offers tracking and disturbance rejection against unknown set-points and/or disturbances. The proposed approach is successfully applied to a highly non-linear, coupled, water-tank process which exhibits both minimum and non-minimum phase characteristics.