Design of PI Controller for Consistency of Stock in Paper Machine Headbox using Particle Swarm Optimisation (PSO)

Abstract

This paper presents a PSO based design of classical Proportional-Integral (PI) controller for consistency of stock, one of the most important papermaking parameters used as a single input single output (SISO) system. Various transfer function models of this parameter reported in the literature have been employed. These transfer functions are actually developed from experimental data from real systems i.e. paper industry. Out of so many tuning techniques for controller settings, the Ziegler-Nichols (ZN), internal model control (IMC) & Tyreus-Luyben (TL) techniques are mostly used in industries. These controller tuning methods usually produce large overshoot, larger settling time slower dynamics and less comparative stability. Therefore, modern heuristics approach such as particle swarm optimization (PSO) is proposed to examine its suitability in enhancing the capability of traditional techniques. For comparison purposes, the simulation with ZN -PI, TL-PI, and IMC-PI are also presented. The analysis reveals that the PSO-based PI approach in most of the situations provides better optimal performance pertaining to overshoot, frequency response & relative stability (gain Margin and phase Margin), and low-performance index values of given process models of stock consistency.