A Novel Self-Adaptive Casting Net-Based Particle Swarm Optimization

Prematurity is a troublesome problem that has to be faced and got rid of by many optimization algorithms, especially the Particle Swarm Optimization (PSO). To combat with prematurity, this paper proposes a self-adaptive casting net mechanism that is able to search global fitness efficiently. To keep diversity of particles, the self-adaptive casting net mechanism tunes parameters dynamically according to the number of iteration. Based on the proposed casting net mechanism, a novel Self-adaptive Casting Net-based Particle Swarm Optimization (SCNPSO) is presented. Experiments were carried out to compare the standard PSO with SCNPSO with various parameters for self-adaptive and different strategies for moving based on benchmark functions of optimization. Experimental results show that SCNPSO outperforms PSO due to adjusting parameters self-adaptively and strategies for moving.