Individualism of particles in particle swarm optimization

Autor: Chen Li, Xiaolin Mao, Kun Miao
Rok vydání: 2019
Předmět:
Zdroj: Applied Soft Computing. 83:105619
ISSN: 1568-4946
Popis: The particle swarm optimization (PSO) method is an effective, nature-inspired, computational algorithm for optimization problems. However, the influence of individuals’ cultural orientations is neglected in particle swarms. Individualist and collectivist orientations have an important influence on optimization. In order to improve the performance of PSO in nature, particularly with respect to the balance of exploitation and exploration in the search process, and in view of the inherent characteristics of particles, a few particles in the swarm was regarded as having an individualistic orientation, which may be beneficial to group creativity from a cultural psychology perspective. The particles holding individualistic orientations were named individualism particles (I-particles). To simulate the divisive and unruly features of I-particles, a random term was introduced to the velocity updating formula to simulate the creative behavior. The experiment was performed with and without I-particles in the PSO algorithm. The presence of I-particles produced better performance in terms of solution accuracy and convergence speed. Furthermore, when added to one of the PSO variants, the I-particles could also contribute to improve the performance of the PSO variant. Further, even for complex CEC2013 benchmark functions, good results were achieved in most problems when I-particles were added to the PSO. Besides, the study is not focused on a PSO variant algorithm, but on the nature of the PSO. Thus, The findings of the nature indicated that I-particles in a particle swarm might be regarded as a supplement to the basic structure of PSO.
Databáze: OpenAIRE
Externí odkaz: