Study On Assessment, Improvement And Application Of Differential Evolution Algorithm

In recent years,the evolution of computing as a new discipline of artificial intelligence is developed very rapidly.Differential evolution algorithm is mainly used for continuous variables or peak multidimensional global optimization problem,it is mainly work steps is basically the same as other techniques,including mutation(Mutation),crossover(Crossover),select(Selection)three operations.The basic idea of the differential evolution algorithm is:in real coding scheme,randomly generated initial population,randomly selected two individuals from the population.The difference between vector is used as a random variations in the third individual source.Weighte difference vector,and then follow some certain rules and sum the third individual.Then produce individual variation,this operation is called mutation.Then mix the parameters of individual variation and the pre-determined target individual,generate test individual,this process is called cross.If the fitness function of the test individual are better than the target individual,then the test individual in the next generation will replace the target individual,or target individual will still be retained,this operation is called choice.In the generations of evolution,each individual will be used as a vector of the target individual.Ultimately retained better individual,out of poor individuals,combined with Darwin's theory of survival of the fittest ideology,guide the search process to approache the global optimal solution.This article describes the background and development as well as the basic principles of Differential Evolution Algorithm,Optimizationand,Evolutionary Algorithm sproposed and ten different mutation operator according to the configuration of difference vector modes to form dozens of strategies.Through experimental analysisand,experimental data obtained use the fitness function to evaluate each policy,select the best fitness function value as the best strategy.