Global Convergence Analysis Of Hybrid Genetic Algorithm

Widely known, Traditional Optimization Method(TOA)[1,2] has had very maturetheory about convergence in research. But TOA are local searching algorithms. So wemust put up the task about global searching Algorithm's research in convergence. Thesewere many result about Genetic Algorithms[3,4,5],including Markov model andVose-Liepins model. Markov model is a very common and very natural method. It hadbe applied this many a day in research. The people find that it has not advantage morethan TOA in function optimization. For example ,weak local searching capacityexistence of prematurity and stochastic ramble and the like, so resulting in weakconvergence of genetic algorithms, it must take a long time to find global solution. Thisshortage baffle the widely application of genetic algorithms. It is a significant taskwhich scholars is groping for how to improve the searching capacity of geneticalgorithm and converging pace to make it be applied in practice. Especially it is a mainmethod of improving genetic algorithm[55,56,57,58,61,62]to messy genetic algorithm withTOA possessing climbing capacity, It has been regarded by many numeric operator. Butthe job centralizes in application of algorithms, the research of theory lag the research ofpractice application greatly. This paper is illumed by reference[61],and commingle genetic algorithm with TOAfrom the angle of quasi-descent algorithm .fusing the advantage of steeply convergenceand highly precession of TOA and global convergence of genetic algorithm. Makingnew hybrid algorithm has the advantage of both. In third chapter, we take theory analysis of global convergence algorithm of thispaper. By the end of paper, we put up several numerical simulation result of testfunction on the hybrid genetic algorithm of this paper .Testing the feasibility ofalgorithm and reliability of global convergence in theory.