Effect of tuning parameters on performance of first-order plus dead-time processes using Generalized Predictive Control

Generalized Predictive Control (GPC) belongs to a family of Model Predictive Controllers, which incorporate the concept of Receding Horizon to predict the future output, optimize a cost function based on the predicted error and provide the optimized control sequence. GPC controller has four tuning parameters, control weight, minimum and maximum prediction horizon and control horizon. The paper addresses the effect of these tuning parameters on system performance in terms of percentage peak overshoot, settling time and rise time. The effect was studied on a stable first-order plus dead-time (FOPDT) system for set-point tracking. Based on the results obtained, a set of thumb rules for GPC tuning are proposed. The rules are not exact quantification of performance criteria in terms of the tuning parameters, but serve as a guideline. Literature suggests that any process can be approximated as an FOPDT process for control purposes, hence the rules can be extended to other process examples.