Predominant Cognitive Learning Particle Swarm Optimization for Global Numerical Optimization

Particle swarm optimization (PSO) has witnessed giant success in problem optimization. Nevertheless, its performance seriously degrades when coping with problems a lot of local optima. To alleviate this issue, paper designs predominant cognitive learning particle (PCLPSO) method to effectively tackle complicated problems. Specifically, for each particle, new promising exemplar is constructed by letting personal best position cognitively learn from better experience randomly selected those others based on novel strategy. As result, different particles preserve guiding exemplars. In way, the effectiveness and diversity are expectedly improved. eliminate dilemma that PCLPSO sensitive involved parameters, we propose dynamic adjustment strategies, so parameter settings, which further beneficial promote particles. With above techniques, proposed could compromise search intensification diversification good way complex solution space properly achieve satisfactory performance. Comprehensive experiments conducted commonly adopted CEC 2017 benchmark function set testify devised PCLPSO. Experimental results show obtains considerably competitive or even much more than several representative state-of-the-art peer methods.