Title: Evolutionary optimisation-based design of 2-DOF observer controller for integral process with dead time: a comparative study

Authors: Ashu Ahuja; Shiv Narayan; Jagdish Kumar

Addresses: Electrical Engineering Department, M.M. University, Mullana, India ' Electrical Engineering Department, PEC University, Chandigarh, India. ' Electrical Engineering Department, PEC University, Chandigarh, India

Abstract: This paper presents a design methodology for two degrees of freedom (2-DOF) observer-based controller for the integral processes with dead time. The 2-DOF controller structure facilitates independent design for feedback properties and set point tracking. The feedback controller aims to achieve stability, robustness and disturbance attenuation whereas prefilter design produces satisfactory set point response. In the approach presented here, the feedback controller and prefilter designs are posed as optimisation problems which are solved using various optimisation techniques such as particle swarm optimisation (PSO), genetic algorithm (GA) and differential evolution (DE). Feedback controller is designed by minimising H∞ norm of the mixed sensitivity and feedforward controller is designed by minimising integral square error (ISE) using model matching problem. The above said three optimisation techniques are compared for both performance indices. It has been shown that GA and PSO optimised controllers have better robustness and set point tracking than DE optimised controller.

Keywords: 2DOF controller; dead time; coprime factorisation; H-infinity control; particle swarm optimisation; PSO; genetic algorithms; GAs; differential evolution; evolutionary optimisation; controller design; feedback control; set point tracking; stability; robustness; disturbance attenuation; prefilter design; robust control.

DOI: 10.1504/IJAAC.2015.068042

International Journal of Automation and Control, 2015 Vol.9 No.1, pp.1 - 23

Received: 09 Jun 2014
Accepted: 15 Jul 2014
Published online: 15 Mar 2015 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article