Differential Evolution Algorithm Research Based On The Combination Operation Of Mutation Strategy And Parameters

Evolution algorithm inspired by biology, especially the individuals in the process of biological evolution are organized in order to adapt to the change of the surrounding environment. When using evolution algorithm, it is always need to clear how individuals will generate new individuals, and some parameters will be need in the process, the parameters could decide the individual quality and search efficiency to a large degree.Differential evolution algorithm is easy to understand and use, one of intelligent evolution agorithms with good performance, it is not only has fast convergence speed, but also has simple operation, so differential evolution algorithm research has become one of the important topics in the field of optimization method. Based on the research of the existing differential evolution algorithm in this paper, insight into the algorithm performance influence of each mutation strategy and the parameters, in view of the current existing algorithm with random combination of mutation strategy and parameters, to each accroding to his need not accroding to the mutation strategy and parameters properly, so the analysis and study of the mutation strategy and parameters influence to the differential evolution algorithm, then came up with a optimization algorithm based on the combination operation of mutation strategy and parameters.Firstly, this paper analyzes and studys the differential evolution algorithm process,application and the mutation strategy, parameters, masters the influence of each item, then clears the problems. In view of the random combination of mutation strategy and parameters in some current algorithms, adapting the operation of cross-combination,raising a composite differential evolution algorithm based on cross-combination of mutation strategy and parameters, searching for better individuals from a larger range,providing more choices for the next iteration.Secondly, this paper analyzes the characteristics of each optimization problem, finds the blind match phenomenon of mutation strategy and parameters, that is lacks of pertinence. In view of the problem, through the matching degree analysis of optimization and mutation strategy and parameters, adapting the method of classification, and the optimization problems are divided into two categories: the basic combinating problemsand the complex transformation problems, then putting forward to a hybrid differential evolution algorithm based on the classification.Thirdly, through comprehensive the above two operations: cross-combination and classification, considering the effects of composite of mutation strategy and parameters,classification, adapting the fusion of cross-combination and classification, then a differential evolution algorithm based on the classification and the cross-combination thought has proposed.Finally, through the use of MATLAB as the experiment platform, based on the combination of mutation strategy and parameters of differential evolution algorithm research for validation. From the rate of convergence, function error value, and many other performance evaluation standard, comparing the proposed algorithms and a variety of experimental data of existing algorithms, so as to verify the effectiveness of the proposed algorithms.