基于控制参数双峰分布的小种群差分进化算法

当前差分进化算法研究主要集中在常规种群上，对小种群差分进化（DE）算法的研究较少。小种群差分进化算法因种群规模小，存在多样性降低过快的问题。因此提出一种基于控制参数双峰分布的小种群差分进化算法（BiMDE）。该算法采用基于柯西双峰分布的参数调节机制处理变异缩放因子 F 和交叉概率因子 CR，并对缩放因子 F 进行矢量化设定。将 BiMDE 算法在函数集 CEC2014 上进行测试，并与 5 种最新的小种群差分进化算法进行比较。结果表明，BiMDE 算法在求解精度、收敛速度以及多样性保持上具有较大优势。

Micro Differential Evolution Algorithm Based on Bimodal Distribution of Control Parameters

At present，researchers mainly focus on studying differential evolution（DE）algorithm with large population size，while there are few studies on the DE with small population. Since DE algorithm with small population has the problem of rapid decrease in diversity，a micro differential evolution algorithm based on bimodal distribution of control parameters（BiMDE）is proposed in this paper. In the BiMDE algorithm，bimodal distribution adjustment mechanism based on Cauchy distribution is adopted for two control parameters，i.e.，scaling factor F and cross probability factor CR，and vectorization setting is also used for scaling factor F. The proposed BiMDE algorithm is evaluated on the 30 CEC2014 benchmark functions，and compared with the four latest micro DE algorithms. The results show that the BiMDE algorithm has much enhanced performance in terms of solution accuracy，convergence speed and diversity maintenance