摘要
克隆选择算法被广泛应用到各个领域,为解决DeCastro克隆选择算法中存在的一些问题:需要根据人为经验确定种群规模的大小、种群训练的时间比较长、多峰搜索能力相对较弱,对其进行进一步的改进,运用新的克隆选择、克隆变异和最佳亲和度,并引入了抗体抑制操作,可动态确定种群大小,使算法具有较强的全局和局部搜索能力,同时也可以搜索到全局最优点和尽可能多的局部极值点。简单仿真实验结果表明,该算法的平均运行时间和找到峰值点个数都明显优于DeCastro克隆选择算法。
Clone selection algorithm is widely applied to various fields, in order to solve the existed problems of DeCastro clone selection algorithm that are the population size determined by the experience, relatively long population training time, weaker multi-peaks search cap ability, based on the analysis of clone selection algorithm made a further improvement, used new clone selection operation, clone mutation operation and the best affinity, and adopted the antibody suppression operation. The algorithm can dynamically determine the population size and has strong abilities of global and local search, also can search for global optimum and so many local minimum points. Simulation results show that the algorithm found the average running time and numbers of the peaks are much better than DeCastro clone selection algorithm.
出处
《计算机技术与发展》
2012年第5期101-104,共4页
Computer Technology and Development
基金
国家自然科学基金资助项目(70871067)
辽宁省自然科学基金资助项目(20072207)
关键词
人工免疫
克隆选择算法
亲和度
多峰搜索
artificial immune
clone selection algorithm
affinity
multi-peaks searching
