Adaptive Hybrid Differential Evolution Algorithm and Its Application in Fuzzy Clustering

To improve the globe searching ability of differential evolution algorithm (DE), an adaptive hybrid differential evolution algorithm (AHDE) is proposed. The cross operator of the proposed algorithm is adjusted according to the computation process to enhance the globe convergence ability of the algorithm. Simulated annealing (SA) is adopted for its strong local search ability to overcome the premature convergence of DE. The test results of Several Benchmark functions show that AHDE can avoid premature effectively and its globe convergence ability is better than that of DE. A new fuzzy clustering method combined AHDE with Fuzzy C-Mean algorithm (FCM) is presented and experiment results show that the clustering method presented can avoid the limitation of converging to the local optimal point of FCM and the clustering results obtained are more rational than those from FCM.