Controller tuning, stiction and multi-objective optimization: an exploratory analysis

Abstract

This paper addresses the problem of adjusting PI controllers in control loops subject to stiction. Given their success in similar environments, multi-objective optimization techniques could be an alternative for adjusting these controllers. As no previous works have explored this approach, several questions arise. Therefore this paper presents a preliminary study to analyze the Pareto front of the controllers adjusted for stiction environments. A Pareto front approximation is obtained for the case without stiction, and its performance is evaluated for increasing levels of stiction. A grid search approach is used to find the Pareto-optimal values of Kp and Ki, and the resulting Pareto front approximations for each combination of stiction and objective function for multi-criteria analysis. The results of this study can provide useful insights for designing control strategies in systems with stiction.