摘要
考虑人工根系算法(AR)对目标函数过于敏感这一特性,以及由此带来的算法早熟和算法失效这两种情况,提出一种生长区域因子基于柯西列收敛的人工根系算法(CAR)。首先按照基本人工根系算法的寻优方式对各根系进行适应值评估,确定其生发新根能力;其次在一次生发新根的过程中,控制生长区域因子按柯西列收敛,使算法逐步从全局搜索收敛到局部搜索。将该算法与AR算法进行比较,数值仿真结果表明对某些函数极值问题改进算法提高了全局搜索能力和收敛精度,改善了优化性能。
The artificial root（ AR） algorithm is sensitive to the objective function,and AR results in premature and failure in optimization. An artificial root algorithm whose modified growth factor is based on the convergent Cauchy-sequence（ CAR） is proposed. First,this paper evaluates the fitness of the roots and calculates their abilities of root-growing according to the optimization method of the basic artificial root algorithm; then,during the process of a root-growing,the growth factor is selected as a Cauchy-sequence. Finally,the proposed algorithm is compared with AR algorithm. The experiment results show that,for some function extreme problem,CAR can effectively improve the global searching ability and enhance the accuracy of convergence.
出处
《信息技术》
2016年第5期175-178,182,共5页
Information Technology
关键词
人工根系算法
柯西列
生长区域因子
智能优化
artificial root algorithm（AR）
Cauchy-sequence
growth factor
intelligence optimization
