An Expert System for Intelligent Selection of Proper Particle Swarm Optimization Variants

Abstract

Regarding the large number of developed Particle Swarm Optimization (PSO) algorithms and the various applications for which PSO has been used, selecting the most suitable variant of PSO for solving a particular optimization problem is a challenge for most researchers. In this paper, using a comprehensive survey and taxonomy on different types of PSO, an Expert System (ES) is designed to identify the most proper PSO for solving different optimization problems. Algorithms are classified according to aspects like particle, variable, process, and swarm. After integrating different acquirable information and forming the knowledge base of the ES consisting 100 rules, the system is able to logically evaluate all the algorithms and report the most appropriate PSO-based approach based on interactions with users, referral to knowledge base and necessary inferences via user interface. In order to examine the validity and efficiency of the system, a comparison is made between the system outputs against the algorithms proposed by newly published articles. The result of this comparison showed that the proposed ES can be considered as a proper tool for finding an appropriate PSO variant that matches the application under consideration.

Keywords

• Particle swarm optimization
• Taxonomy
• PSO variants
• Expert system
• Knowledge base