Improved differential evolution algorithms
Chengfo Sun, Haiyan Zhou, Liqing Chen · 2012
Before improving the differential evolution (DE), the premature convergence feature of the differential evolution must be analyzed, which demonstrates that the differential evolution is not able to guarantee the global convergence. In order to improve the searching ability of differential evolution, two modified differential evolution are introduced by merging the mechanisms of quadratic approximation, Gaussian disturbing, immune theory and differential evolution. First, the simplified quadratic approximation is employed to improve the performance of the Immune Self-adaptive Differential Evolution and the novel algorithm is named Quadratic Approximation based Immune Self-adaptive Differential Evolution. Besides, the Gaussian disturbing is introduced into the framework of ISDE to improve the variety of the individual and the proposed algorithm is called Gaussian Disturbing based Immune Self-adaptive Differential Evolution. The performance of the proposed algorithms is tested by the benchmark problems and compared with original DE and ISDE. Both of the proposed algorithms outperform the compared algorithms. This electronic document is a "live" template.