Particle swarm optimisation based on self-organising topology driven by fitness with different links removing strategies. (1st January 2012)

• Record Type: Journal Article
• Authors: Mo, Simin; Zeng, Jianchao
• Is Part Of: International journal of innovative computing and applications. Volume 4:Number 2(2012)
• Page Start: 119
• Page End: 132
• Publication Date: 2012-01-01
• Subjects: particle swarm optimisation -- PSO -- self-organising topology driven by fitness -- SOTDF -- links removing strategies; Evolutionary computation -- Periodicals; Neural networks (Computer science) -- Periodicals; Genetic programming (Computer science) -- Periodicals; Biologically-inspired computing -- Periodicals; Swarm intelligence -- Periodicals; Quantum computers -- Periodicals; 006.3
• Languages: English
• ISSNs: 1751-648X

Abstract

In this paper, a novel particle swarm optimisation (PSO) based on self-organising topology driven by fitness (PSO-SOTDF) is proposed. The topology will be gradually generated as the construction process and the optimisation process progress synchronously. And the construction process of topology involves operations of adding and removing links under invariable network sizes. Further, due to the operation of removing links influencing topology characteristics heavily, three kinds of links removing strategies are designed, which are referred to as Removal Strategy I, II and III. To obtain deep insights, the PSO-SOTDF with Removal Strategy I, II and III are used to solve benchmarks. Simulation results show that Removal Strategy I is more effective than other links removing strategies. In addition, the performance of PSO-SOTDF with Removal Strategy I is compared with other variants of PSO. Simulation results indicate that PSO-SOTDF with Removal Strategy I is competitive.