Analysis And Application Of Genetic Algorithms Population Transition Phenomenon

Optimization problems are closely linked to people’s daily lives. From the choice of pathof personal going out, the arrangement of working time and content, to the design productionof products and making the strategic decision, all of this can be found its shadow faintly.Withthe increasing complexity of optimization problems, the effectiveness of the traditionaloptimization strategies that can be achieved is extremely limited, which need urgently for anew algorithm to improve the optimization performance.In the recent decades, with the rapiddevelopment of computational intelligence, some of intelligence optimization algorithms thatare distinct difference from the principles of the classical mathematical programming haveemerged. Among these, the genetic algorithms have excellent global search capability,implicit parallelism and simple operator, so that they have widely used in many fields.However, the premature convergence of genetic algorithms make a great impact of theperformance of algorithms, and make a serious impediment to further promote geneticalgorithms.Combined with the research experience of early scholars, this paper uses SGA as startingpoint, and find a more special populations transition effects in genetic algorithms of lowmutation rate when do the research of SGA.After do a depth analysis of population transitioneffects and the comparison of a large number of experiments, the author draws conclusionsthat the performance of SGA in low mutation rate is superior to the elitist replacement GA. Inorder to better apply the population transition effects, this paper combining with the conceptof field leaders and the population expulsion in the nature, accelerate the population transitionin the low mutation rate, and then proposes an improvement genetic algorithm based onpopulation transition and regional expulsion (TEGA).Finally, this paper makes comparisons of TEGA, general genetic algorithm, randomreboot genetic algorithm in many aspects. From the result, it proves that using TEGA has agood search capacity of global optimization, which can avoid the premature convergence ofgenetic algorithm.