新型紧致遗传算法及其性能分析  

Novel Compact Genetic Algorithm and Its Performance Analysis

作　　者：彭军[1] 刘振[1] 徐学文[1] 

机构地区：[1]海军航空工程学院接改装训练大队,山东烟台264001

出　　处：《计算机技术与发展》2015年第11期120-124,共5页Computer Technology and Development

基　　金：国家自然科学基金资助项目(61174031;60674090)

摘　　要：针对传统紧致遗传算法收敛速度过慢及容易陷入局部极值的情况,对分布估计算法中的紧致遗传算法进行了研究,提出了一种新型的混合紧致遗传算法,系统地阐述了该算法的流程和进化机制。首先设置主种群和辅种群,通过多种群并行进化设置来加速收敛速度,并通过多个概率向量来控制算法的进化过程,当满足一定进化条件后,在主种群内进行免疫接种,以增加优良个体存活的概率。设置主种群和辅种群的自适应模式交流策略,增加种群多样性,避免过早收敛。对提出的算法在收敛性以及收敛速度上进行了理论分析,证明了算法满足收敛条件,能够确保收敛,并给出了算法的收敛时间的估计。最后利用基准函数进行了函数仿真分析,结果充分验证了所提算法的正确性。Aiming at the slow convergence speed and weak convergence performance for the compact genetic algorithm,a novel compact genetic algorithm is proposed in this paper through research of the compact genetic algorithm in distributed evalution algorithm, the procedure and evolution mechanism is also given. Firstly, setting up primary population and secondary population, the algorithm can accelerate the convergence speed through parallel evolution in multipe population and control the evolution process by various probability vector. In the first primary population, immune vaccination is used to increase the probability of better individual, the first primary and the second population can exchange with each other adaptively in order to prevent premature and enhance the diversity. The convergence and convergence speed of the algorithm is analyzed in theory, which proves that the algorithm can converge, and the convergence time is also estimated in this paper. The simulation results of classic function prove the correctness of the algorithm.

关 键 词：紧致遗传算法 概率向量 收敛性 函数仿真 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]