Min-max model predictive control of a laboratory plant

This work describes the application of a Min-Max predictive controller to a control laboratory plant. Min-max formulations of Model Based Predictive Control (MBPC) are one of the possible approaches in the literature to deal with the control of plants subject to bounded uncertainties. One of the drawbacks of Min-Max MBPC is the amount of calculation required to find a control sequence. The controller used in this paper is relatively efficient numerically: the control sequence is calculated solving a linear programming problem with a reasonable number of constraints. This allows the calculation time to be small enough to apply the controller to reasonably fast real systems. The Min-Max MBPC was tested on-line in a two-tank laboratory plant under different working situations to check the robustness, stability and general performance of the designed controller. The results are quite encouraging.