Modification of particle swarm optimization with human simulated property

• Published in: Neurocomputing (Amsterdam) Vol. 153; pp. 319 - 331
• Main Authors: Tang, Ruo-Li, Fang, Yan-Jun
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 04.04.2015
• Subjects: Algorithms; Cognition; Computer simulation; Convergence; Deep extended memory; Extended memory depth; Forgetting factor; Forgetting function; Human; Human brain simulated property; Optimization; Particle swarm optimization; Strategy; Swarm intelligence; Forgetting function; Extended memory depth; Forgetting factor; Deep extended memory; Human brain simulated property; Particle swarm optimization
• ISSN: 0925-2312; 1872-8286
• DOI: 10.1016/j.neucom.2014.11.015

This study proposes the Human-brain Simulated Particle Swarm Optimization (HSPSO) and its Further Improved algorithm (HSPSO-FI), in order to improve the evolutionary performance of PSO and PSO-variants. Inspired by human simulated properties, modifications proposed in this article are as follows: Firstly, accumulating historical cognition by the deep extended memory; Secondly, introducing a new learning method of cognition and a new updating strategy of velocity; Finally, defining and analyzing the "forgetting function", "forgetting factor" and "extended memory depth". Evidence from simulations indicates that the extended memory and new velocity choosing and updating strategies can give the moving direction to each particle more intelligently and help them avoid trapping into local optimum effectively, and the novel algorithms have a better performance in convergence speed and optimization accuracy on the test of several benchmark functions.