Particle Swarm Optimization: A Survey

Swarm Intelligence (SI) describes the evolving collective intelligence of population/groups of autonomous agents with a low level of intelligence. Particle Swarm Optimization (PSO) is an evolutionary algorithm inspired by animal social behaviour. PSO achieves performance by iteratively directing its particles toward the optimum using its social and cognitive components. Various modifications have been applied to PSO focusing on addressing a variety of methods for adjusting PSO's parameters (i.e., parameter adjustment), social interaction of the particles (i.e., neighbourhood topology) and ways to address the search objectives (i.e., sub-swarm topology). The PSO approach can easily fit in different search optimization categories such as Self Learning, Unsupervised Learning, Stochastic Search, Population- based, and Behaviour-based Search. This chapter addresses these principal aspects of PSO. In addition, conventional and Basic PSO are introduced and their shortcomings are discussed. Later on, various suggestions and modifications proposed by literature are introduced and discussed.