基于引入强制变异的改进遗传算法性能研究【毕业论文】.doc

本科毕业论文 （20 届） 基于引入强制变异的改进遗传算法性能研究 专业：计算机科学与技术 摘 要 遗传算法GA(Genetic Algorithm)是 Performance Research Based on Improved GA With Compulsory Mutation Abstract Genetic Algorithm（GA）developed in recent decades , is an Artificial Intelligence technology simulating natural biological process and mechanism to solve extremum problem. It was proposed by professor Holland from University of Michigan , come from the research of Artificially adaptive sysetem. It using natural seletion and mutation mechanism to creat a global optimization adaptive probability search Algorithm. When running GA , there are contradictions about convergence rate and global convergence between selection operator and crossover and mutation operator. In the eraly evolution, Individuals of high fitness will control the process of evolution quickly , lead to premature group , and in the late evolution ,little difference of individuals lead to slow convergence rate. Crossing and mutation are effective methods to select new region of solution space. But operations of Crossing and mutation rely on experience,and different problems with different situations. High Crossover probability will eliminate high fitness individuals quickly ; Low Crossover probability will stop the searching.Similarly mutation probability has the same problem. So we have to inprove the simple type of genetic algorithm. To solve this problem, Srivivas and his parteners proposed the idea of Adaptive .The main idea of it is to change cross probability and mutation probability according to the relationship of max fitness and average fitness. But normal adaptive GA still has a shortage, it is easy to get the local best answer. To solve these problems a proved GA with compulsory mutation , adaptive corssing and variation , reservations of best solutions is proposed in this paper. The principle of the new algorithm is as follows : in normal situations we use adaptive parameters to handle with crossing and mutation probability, and do single