Model predictive control of quadruple tanks system

In this paper, the application of offset-free model predictive control to a quadruple tanks system is addressed. The main goal is to eliminate steady state tracking error which is produced by model mismatch. The model mismatch between process and model used by model predictive controller is due to parameter uncertainty. It is shown how this uncertainty can be modeled by state disturbances. In this method, the disturbances are augmented to process states and an observer is designed to estimate them. The results show how this method increases the robustness of the model predictive controller.