An improved constriction factor particle swarm optimization algorithm to overcome the local optimum

In order to solve the problems of low efficiency and premature convergence, an improved constriction factor particle swarm optimization algorithm, abbreviated to ICFPSO, was proposed in this paper. Position and speed factors were introduced as two new parameters to judge the stagnation of particles. For each individual, when the distance between its position and the current global optimum was less than the pre-set position factor and its velocity less than the pre-set speed factor, then this particle was thought to fall into local optimum. Meanwhile, the position of such particle was re-initialized in the whole solution space. The population diversity of the swarm was enhanced significantly by this method. Three typical multimodal functions were used to verify the performance of ICFPSO. The simulation results show that the improved algorithm had better convergence accuracy and effectively avoided falling into local optimum.