基于多样性保持的遗传算法性能研究【毕业论文】.docVIP

本科毕业论文 （20 届） 基于多样性保持的遗传算法性能研究 专业：计算机科学与技术 摘 要 遗传算法(Genetic Algorithm，简称GA)是由密切根大学Holland等创立的，来源于进化论和群体遗传学，原先用于模拟自然系统的自适应现象，后来被引向了广泛的工程问题。遗传算法发展成一种自适应启发式概率性迭代式全局搜索算法，以其解决不同非线性问题的鲁棒性、全局最优性、不依赖于问题模型的特性、可并行性及高效率，具有独特的吸引力，正引起越来越大的研究及应用热潮[1]。 但是，尽管遗传算法在很多领域具有广泛的应用，但它仍存在早熟收敛问题，为克服标准遗传算法存在的早熟收敛问题，提高优化效果，本文从维持种群的多样性出发， 首先引入了一种基于适应度方差的个体多样性度量指标，并利用指标对交叉概率进行自适应调整，以维持种群中的个体多样性。其次，通过计算各基因座上等位基因的基因多样性，自适应调整相应基因座上等位基因的变异概率，以增强基因多样性、 抑制种群中的有效基因缺失，从而提出了一种新的基于多样性保持的自适应遗传算法 (Diversity Maintaining Adap tive Genetic Algorithm，DMAGA)。 多样性保持的遗传算法是一个要求综合考虑个体多样性和基因多样性克服标准遗传算法存在的早熟收敛问题的算法。遗传算法在组合优化问题上的应用是遗传算法的主要应用之一。本文讲述的单背包问题是一个典型的组合优化问题，根据多样性遗传算法的特点，我们可以很好的将此算法应用于背包问题。因此在组合优化领域上，多样性保持的遗传算法得到广泛的应用与研究。仿真结果表明，算法具有不易陷入局部最优，解的精度较高，达到优化效果的要求。 【关键词】遗传算法，种群多样性，有效基因，早熟收敛，组合优化  Genetic Algorithm Performance Research Based on Diversity Maintaining Abstract Genetic Algorithms (GA) is established by the University of Michigan, Holland, etc., and from the theory of evolution and population genetics, the original adaptive system to simulate natural phenomena, was later to lead a wide range of engineering problems. An adaptive genetic algorithm into the probability of the iterative heuristic global search algorithm, to solve various nonlinear problems in its robustness, global optimality, the model does not depend on the problem characteristics and high efficiency of parallel, has a unique attraction, is causing growing research and application of wave. But, while many areas of genetic algorithm in a wide range of applications, but it is still premature convergence, the standard genetic algorithm to overcome the existing problem of premature convergence and improve the optimization results, the paper starting from the maintenance of population diversity, first introduced a variance of individual fitness based diversity measures, and use of indicators of adaptive crossover probability in order to maintain the population of individual diversity. Secondly, by calculating the locus allele diversity, adap