摘要
小生境遗传算法与遗传算法相比,在求解多峰函数等最优化问题上具有显著的优势,但是小生境距离参数的确定缺乏理论依据,限制了小生境遗传算法的应用。该文提出了一种求解小生境之间距离参数的新方法——基于遗传算法进化优化小生境距离参数。根据多峰目标函数的具体情况,应用遗传算法随机寻优得到若干个最优值,由这些最优值的最小欧氏距离指导小生境距离参数的取值。依据此方法确定小生境之间的距离参数,应用小生境遗传算法成功求解了Shubert多峰函数的所有全局最优值以及六峰值驼背数Back Function的所有局部极小值。
Niche genetic algorithm is superior to genetic in multiple hump function optimization. However, there is lack of theory to determine the parameter of niche distance, so the algorithm＇s application is limited. This paper presents a new approach to determine the niche distance parameter, which is based on genetic algorithm. According to the information of multiple hump object function, genetic algorithm is used to seek several global optimums, whose Euclidean distances are calculated. The niche distance is determined by the minimal Euclidean distance. This approach is successfully used in Shubert function optimization and six-hump camel back function optimization.
出处
《计算机工程》
EI
CAS
CSCD
北大核心
2006年第13期206-208,共3页
Computer Engineering
基金
国家自然科学基金资助项目(60475002)
关键词
遗传算法
小生境
多峰函数最优化
Genetic algorithm
Niche
Multiple hump function optimization
