Authors: Xingjuan Cai
Addresses: Complex System and Computational Intelligence Laboratory, Taiyuan University of Science and Technology, 030024, China
Abstract: Dispersed particle swarm optimisation (DPSO) is a new variant of particle swarm optimisation by introducing a distributed social parameter setting. However, this individual strategy for each particle is setting to a linear manner. Due to the complex nature of engineering optimisation tasks, a predefined linear manner cannot always meet the optimisation requirements. Therefore, in this paper, four non-linear selection strategies are introduced to DPSO to further improve the performance. Four famous numerical benchmarks are used to assess the new algorithm. Simulation results show the exponential curve strategy can achieve best performance than three other non-linear strategies though the improvements relied on the test functions significantly. Furthermore, when compared with three other variants of particle swarm optimisation, the exponential curve strategy also provides the best performance.
Keywords: social parameter; particle swarm optimisation; dispersed PSO; nonlinear selection; exponential curve; simulation.
International Journal of Modelling, Identification and Control, 2009 Vol.8 No.4, pp.301 - 308
Available online: 09 Dec 2009 *Full-text access for editors Access for subscribers Purchase this article Comment on this article