Improved differential evolution with dynamic mutation parameters

Yifeng Lin, Yuer Yang, Yinyan Zhang

Abstract: Differential evolution (DE) algorithms tend to be limited to local optimization when solving complex optimization problems. Different iteration schemes lead to different convergence speeds. In this paper, we mainly use the dynamic mutation parameter \(\text {FS}\) to improve the DE algorithm. Based on two ideas, a total of seven DE schemes are proposed to optimize the DE algorithm. We test the performance of the improved DE scheme on 56 test functions. Experiments show that the improved DE algorithm is better than the baseline DE algorithm in terms of accuracy, convergence and8 convergence speed.