A Hybrid ACO and Nelder-Mead constrained algorithm for controller and anti-windup tuning

Abstract

Nonlinear controller tuning with anti-windup is not trivial since the parameters within the structure are interdependent. A global optimization-based tuning approach can handle this interdependence. A hybrid Ant Colony Optimization (ACO) and Nelder-Mead (NM) algorithm for controller and anti-windup tuning efficiently solves this problem. This paper proposes a new way to constrain ACO-NM algorithm to ensure positivity of key parameters in order to avoid searches in instability or low robustness regions and to improve computation time to reach an optimal solution.