K-means particle swarm optimization with embedded chaotic search for solving multidimensional problems

摘要

The proposed approach inherited the paradigm in particle swarm optimization (PSO) to implement a chaotic search around global best position (gbest) and enhanced by K-means clustering algorithm, named KCPSO. K-means with clustering property in PSO resulted in rapid convergence while chaotic search with ergodicity characteristic in PSO contributed to refine gbest. Experimental results indicated that the proposed KCPSO approach could evidently speed up convergence and successfully solving complex multidimensional problems. Besides, KCPSO was compared with canonical PSO in performance. And, a case study was also employed to demonstrate the validity of the proposed approach.