一种自适应多亲遗传算法及其性能分析  

Adaptive multi-parent genetic algorithm and its performance analysis

作　　者：吴佳英[1] 李平[1] 郑金华[2] 

机构地区：[1]长沙理工大学计算机与通信工程学院,湖南长沙410076 [2]湘潭大学信息工程学院,湖南湘潭411105

出　　处：《系统工程与电子技术》2007年第8期1381-1384,共4页Systems Engineering and Electronics

基　　金：湖南省自然科学基金(05JJ30125);湖南省教育厅基金(06B005)资助课题

摘　　要：定义了一种度量多样性的指标,来自适应地调整基于实数编码的多亲遗传算法中交叉概率和变异概率这两个参数。提出了一种基于实数编码的自适应多亲遗传算法,该算法能自适应地调整其参数,且在求解优化问题的过程中,能克服早熟收敛的现象,提高搜索能力,加速收敛速度。同时将该算法用来求解多峰值测试函数,实验结果验证了算法的有效性。The limitation of genetic algorithms based on binary coding in solving the optimizing problem with continuous parameters is considered. Adopting real coding, an index that can reflect the diversity of population is defined to adjust the probabilities of crossover and mutation in a multi parent genetic algorithm. Thus the adaptive multi parent genetic algorithm based on real coding （AMGARC） is proposed. AMGARC can adjust automatically its main parameters. This algorithm can also overcome premature convergence, improves the optimizing ability and accelerates the convergence speed. Two multimodal functions are optimized with this algorithm. Emulational experimental figures show AMGARC has nice performance.

关 键 词：多亲遗传算法 实数编码 自适应 种群多样性 优化问题 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]