引用本文:巩敦卫, 孙晓燕.用于函数优化的自适应类种子保留遗传算法[J].控制理论与应用,2005,22(5):779~782.[点击复制]
GONG Dun-wei,SUN Xiao-yan.Genetic algorithms with adaptively conserving species seeds for function optimization[J].Control Theory and Technology,2005,22(5):779~782.[点击复制]
用于函数优化的自适应类种子保留遗传算法
Genetic algorithms with adaptively conserving species seeds for function optimization
摘要点击 1201  全文点击 916  投稿时间：2004-05-25  修订日期：2005-01-26
查看全文  查看/发表评论  下载PDF阅读器
DOI编号  
  2005,22(5):779-782
中文关键词  遗传算法  类种子  保留  数值函数优化
英文关键词  genetic algorithms  species seeds  conservation  numerical function optimization
基金项目  国家自然科学基金资助项目(60304016)
作者单位
巩敦卫, 孙晓燕 中国矿业大学信息与电气工程学院,江苏徐州221008 
中文摘要
      类种子保留遗传算法可以较好地处理维持进化种群多样性和保留重要个体的矛盾,但尚无有效方法确定其类控制参数.本文提出一种类控制参数随进化进程自适应变化的策略,其思想是:在进化前期类控制参数较大,将进化种群分成数目较少的粗类;随着进化的进行类控制参数自适应减小,将进化种群分成数目较多的细类.另外,个体的自适应变异充分利用了个体当前状态、本类种子和种群最优种子的信息.将该算法应用于5个基准数值函数优化问题,计算结果验证了本文算法在找到多个极值点的前提下有效地减少了计算量、提高了进化效率.
英文摘要
      Genetic algorithms with conserving species seeds can cope with the conflict between maintaining the diversity of the evolutionary population and conserving the important individuals,but there is no effective way to determine their species control parameter.A strategy for adaptively changing the species control parameter along with evolutionary phase is proposed.The methodology adopted is that the species control parameter is big in prophase so that the evolutionary population is divided into a small number of coarse species.Along with evolution,the species control parameter decreases adaptively so that the evolutionary population is divided into a large number of fine species.Besides,the adaptive mutation operator makes fully use of the information of the current state of being mutated individual,its own species seed and the best seed of the evolutionary population.The algorithms proposed are applied to 5 benchmark problems of numerical function optimization.It is validated from the computational results that the algorithms decrease the computational complexity on the premise of finding multi-optima of the problems being optimized and hence increasing the evolutionary efficiency.