Multi-strategy different dimensional mutation differential evolution algorithm
Peng Xiao, Dexuan Zou, Zhenglong Xia, Xin Shen · AIP conference proceedings · 2019
Aiming at the characteristics of differential evolution (DE) algorithm, such as premature convergence and low convergence accuracy, we propose a multi-strategy different dimensional mutation differential evolution (MDMDE) algorithm. On one hand, the algorithm starts from the mutation dimension, and then proposes a different dimensional strategy. On the other hand, multi-strategy mutation is used in the whole generation cycle of the algorithm. In addition, a modified crossover rate and a dynamic mutation factor are proposed to jump out of the local maximum. Finally, MDMDE is applied to 8 kinds of test functions and compared with other two algorithms. The results show that MDMDE can obtain better optimization results and it exhibits more desirable performance than the other algorithms.