基于邻域搜索的自适应差分进化算法

Adaptive Differential Evolution Algorithm Based on Neighborhood Search

针对传统自适应差分进化算法局部搜索能力薄弱,寻优精度低的缺点,提出一种基于邻域搜索的改进自适应差分进化算法(DADE)。在传统自适应差分进化算法的基础上,通过引入邻域搜索操作,扩大了算法搜索的范围,通过引入高斯扰动来进一步提高算法的收敛速度。在IEEE CEC2013函数集中选取15个基准函数进行实验,与先进的算法变种SinDE、MDE、FWADE,FWA的性能进行比较。实验结果表明,DADE算法在13个函数的误差均值上小于对比算法。

An improved adaptive differential evolution algorithm(DADE) based on neighborhood search was proposed to solve the shortcomings of traditional adaptive differential evolution algorithm with low optimization accuracy and weak local search ability. In the IEEE CEC2013 function set, 15 benchmark functions are selected for experiments, and the performance of advanced algorithm variants SinDE, MDE, FWADE, FWA are compared. Experimental results show that DADE algorithm is smaller than contrast algorithm in the error mean of 13 functions.