Basic Particle Swarm Optimization Based on Reasonable Full Share Information Mechanism
Xiaolin Ren, Jian-Liang Lin, Zhigang Wang · 2007
In order to prevent particle swarm optimization from being trapped in local optima, a new vector called the weighted individual best position of the other particles is introduced. On the one hand, the behavior of each particle is not only influenced by its own best position and the global best position but also by the individual best positions of the others in the swarm. On the other hand, fitness value is used, i.e., each particle weighs the contributions of the others according to their fitness values. So the modified algorithm strengthens cooperation and competition among the particles by making each particle share more useful information of the others. Six benchmark functions are tested and results show that the modified algorithm is more effective than basic particle swarm optimization.