A Hybrid Parallel Approach Based on Chaotic Search and Pattern Search for Multimodal Function Optimization

摘要

A hybrid parallel chaotic search and pattern search (HpCPS) approach for multimodal function optimization is proposed in this paper by hybridizing the parallel chaotic search and parallel pattern search. Chaotic search has good global search ability but poor local search ability; while pattern search method is just in opposite. Both methods are sensitive to the initial points. Parallel chaotic search starts searching from different initial points simultaneously and can effectively reduce the sensitivity of chaotic search. At the same time, chaotic search provides good results as initial points of next pattem search. Based on the results of parallel chaotic search, parallel pattern search is for more approaching the theoretical optimum/optima. The effectiveness of the proposed hybrid approach is validated by numerical examples. The results show that the proposed HPCPS not only enhance the steady of algorithm but also improve the success rate with good precision.