Linear programming predictive control with actuator saturation: Experimental robustness and performance results

Linear programming optimization is widely used in several application areas. One possible application of this optimization technique, which has not received sufficient attention, is process control. In particular, the standard structure of linear programming optimization problems allows formulating optimal predictive control schemes with the additional advantage of facilitating the handling of actuator saturation. Actuator saturation appears in many engineering applications and there are several control approaches dealing with this problem. In this article, a very simple predictive control scheme based on linear programming optimization for actuator saturation control is proposed. The main characteristics of the presented scheme are simplicity and practical robustness/performance tuning parameters. First, the simplest possible output-feedback linear programming predictive controller is proposed as a starting point. Next, through several practical observations inspired in classical control engineering, a series of modifications to the basic controller structure are proposed. These modifications aim at effectively increasing the performance and robustness of the control algorithm. The experimental results confirm that the proposed control algorithm can operate effectively under the presence of perturbations, something which is currently not widely reported in the literature for linear programming actuator saturation predictive control.