基于择优学习策略的差分进化算法

传统的差分进化算法在个体变异方面只是利用了随机个体和最优个体的信息.由于选用个体的随机性,导致其搜索效率比较低并且有可能找不到最优解,为此,提出了基于择优学习策略的差分进化算法.该算法选择性地利用种群中比较优秀的个体的信息,克服种群进化过程中的盲目性,增强了搜索能力.通过对多个具有不同特性的标准测试函数进行测试研究,结果表明该方法可以明显减少迭代次数,提高计算效率.

Perferred Learning Based Differential Evolution Algorithm

Traditional differential evolution (DE) algorithm only abstracts information of random and the best individual. Randomness may result in a low searching efficiency, and even cannot find the best solution. A perferred learning based DE algorithm is proposed to solve this tough problem. The algorithm selectively uses the well behaved individual’s information and overcomes the blindness in the evolving process to enhance the searching ability. After different kind benchmark functions are investigated, the results reveal that the number of iterations can be clearly reduced and the calculation efficiency can be improved.