并行遗传算法的应用.docVIP

廊坊师范学院 本科生毕业论文 题 目：并行遗传算法的应用 学生姓名：李金城 导师姓名：范利强 院 别：数学与信息科学学院 系　　别：信息与计算科学系 专 业：信息与计算科学 年 级：2008级本科1班 学 号：08040341013 完成日期 2012年 4月 25日 廊坊师范学院本科生毕业论文 论文题目：并行遗传算法的应用 论文摘要：遗传算法是模拟达尔文的遗传选择和自然淘汰的生物进化过程的计算模型．传统的遗传算法虽然具有隐含的并行性,但目前大多为串行遗传算法．串行遗传算法在解决一些实际问题时,由于需要较多的个体数量和大量的计算,使得进化过程比较缓慢,难以达到实时的要求．因此并行遗传算法就受到了较大的重视,并且已经成为目前遗传算法研究的主要课题．遗传算法与并行计算机相结合,能把并行机的高速性和遗传算法固有的并行性两者的长处彼此结合起来． 本文举例了一种遗传算法的最优双阈值搜索算法．将种群个体设计成一行两列的向量,以类间方差比作为个体的适应度,生成若干初始子种群进行并行计算；同时,将每代的最大适应度个体直接复制进入下一代,适当增大变异概率,在种群的多样性环境下实施有条件的最佳保留策略,保证算法收敛于最优解．实验结果表明,该算法具有良好的分割效果和较高的运算效率． 关键词： 遗传算法；并行遗传算法；图像分割 Title： The Design and Parallel Realization of Genetic Algorithm Abstract：Genetic Algorithm is the computation model which simulating the biological evolution process of Darwin’s heredity choice and the natural selection． Although the traditional genetic algorithm has the concealment parallelism, but realized the method in essentially still was actually serial． When this kind of serial genetic algorithm solves some actual problems, it will need lots of individuals and computations, and this will causes the evolution process to be so slow that difficult to meets the real-time requirements． Therefore, the Parallel Genetic Algorithm has received a big value, and already becomes the main topic at genetic algorithm research． With the PGA, you can unites the parallel machine’s high speed with genetic algorithm’s inherent parallelism． Each individual in thepopulation is designed as a vector of one row and two columns, in which each element is encodedto binary and using variance ratio between clusters as its fitness value, and several sub-populationsare generated and calculated parallelly． Meanwhile, in the genetic operations, the maximum fitnessindividual in each population is reproduce into the next generation and the mutation factor isincreased properly．