An Improved Particle Swarm Optimization (IPSO) Approach for Identification and Control of Stable and Unstable Systems

Abstract

In this paper, an Improved Particle Swarm Optimization (IPSO) technique is generalized to identify and control four systems of different types of behaviors. This was possible thanks to the use of a new initialization strategy of partitioning of particles, which helps PSO to converge faster to the correct region in the research space. The choice of an enhanced fitness function consisting of the weighted sum of the objectives gives better performances compared to those found using four other commonly used performance indices (ISE, IAE, ITAE, and ITSE). The validity of the model chosen for identifying these four types of behaviors is proved, and the control of these systems using IPSO and many conventional optimization methods such as Ziegler-Nichols, Graham-Lathrop, and Reference Model has been compared and confirmed that IPSO generates a high-quality solution with a short calculation time and a stable convergence feature. Moreover, results confirmed that the IPSO optimized PID is the best as it has good performance and good robustness and it is insensitive to perturbations.