摘要
对用于多峰值函数的人工免疫网络算法进行了改进并分析了其特性。首先在给出基于人工免疫网络的多峰值函数优化算法及其流程的基础上,提出了一种克服早熟现象的改进方案;然后通过与克隆选择算法的数值对比实验,对改进后算法的计算结果加以分析,验证了该算法用于求解多峰值函数优化的有效性;最后重点讨论了算法主要参数对其求解性能的影响,得到了若干参考性结论,可为人工免疫网络计算提供指导。
An improved artificial immune network algorithm for multimodal function optimization along with its characteristic analysis is presented in this paper. Firstly, an artificial immune network algorithm for multimodal function optimization (opt-aiNet) as well as its implementation process is introduced. Based on that we proposed a revised version to overcome the premature phenomenon. Then performance of the improved algorithm is compared with CLONALG, a typical clonal selection algorithm, through some numerical experiments, and its advantages as well as the effectiveness in multimodal function optimization are verified. Finally, the influence of parameters on performance of the improved algorithm is discussed in detail and some valuable conclusions are reached, which can be used to guide artificial immune network computing.
出处
《模式识别与人工智能》
EI
CSCD
北大核心
2005年第1期17-24,共8页
Pattern Recognition and Artificial Intelligence
基金
高等学校博士点基金(No.20030487054)