Studies Of Adaptive Genetic Algorithm On Parameters' Participation In Evolution

Human increasingly adept at studying and simulating life phenomenon and law of nature to get knowledge, then evolve and filter them into other fields successfully. Genetic Algorithm is a high efficient, parallel and global searching algorithm, which was learned from natural selection and genetic mechanism, and was produced from various subjects' integrating and infiltrating with each other. The application of standard genetic algorithm is the simplest part of GA and the basis for other high-level genetic algorithm.Adaptive Genetic Algorithm is one of the improvements of SGA, it is obviously that AGA enhances the speed of convergence, and has excellent performance in the various application fields of genetic algorithms. However, some experimental data show that adaptive genetic algorithms still has some problems, suck as premature convergence and the lack of local searching capability. In recent years with the continuous development of Genetic algorithms and further expand of its application fields, the performance of adaptive genetic algorithm is higher required. The main research work is proposed the new adaptive strategy of parameters that is aimed at the existing problems of adaptive genetic algorithm, and combined with applications of genetic algorithm performance on the new requirements.This paper introduces the biological basis of genetic algorithms, development process, explains the study background, purpose and expected results. Then, we introduce the genetic algorithm and the mathematical theory of genetic algorithm and adaptive genetic algorithm and genetic algorithm to analyze the existing problems in a number of improved algorithms. This new strategy is using the optimization of genetic algorithms to be able to optimize the parameters of genetic algorithm dynamically. New strategy integrates the parameter information into the individual code, and it's involved in the whole process of crossover and mutation operation, so in the process of evolution the new parameters always being produced. Population diversity, convergence rate and individual fitness, and many other factors are considered as one part of the rule of parameter optimization, and the population diversity is measured by the concept of population entropy in this new strategy. The application of a new individual classification rule is used in the realization of the definition of population entropy, and the improvement of the crossover operator and mutation operator is based on the classification rules. Finally, there are abundant experimental data can be used to prove the conclusion that: the new algorithm not only can increase the convergence speed of genetic algorithm, but also can enhance the performance of the global convergence of genetic algorithm to obtain a satisfactory global optimum.