A Random Factor Extension on the PSO Algorithm

The particle of the PSO algorithm is lack ofparticle diversity, and this makes itself fall into thelocal extremum, which leads to the premature. Basedon a large number of experimental analysis, this paperindicated that the reason that lead to the lack of thePSO algorithm´s particle diversity is that PSOalgorithm´s rapid convergence and the smallprobability of the particle speed divergence.According to this, this paper comes up with analgorithm that extend the random factor named RPSOalgorithm from (0, 1) to (-1, 1), which makes theparticle velocity has divergence probability and thenincreases the diversity of the particle speed and so thatincrease the ability to get rid of the local extremum.We made two emulation experiments about multipleextremum constrained optimization problems,experimental results show that, this algorithm is verypossible to jump out of the extremum, and effective toavoid the premature convergence, the result afteroptimizing is obviously better than before it.