Research On Differential Evolution Algorithm Mutation Operators Based On Population Sorting Strategy

Since evolutionary algorithms are self-adaptive, auto-optimizing, universal applicability, it is reasonable and available to solve the complex optimization problems in the science and engineering by using evolutionary algorithms.Differential evolution(DE) algorithm is a population-based evolutionary algorithm. The advantages of DE are its easiness of use, simple structure, efficacy and high robustness. Therefore, it has been widely concerned by researchers. However, the traditional differential evolution algorithm is weak of exploitation, which directly leading to the slow rate of convergence rate in the later evolutionary stage.The salient feature of DE lies in its mutation mechanism. Furthermore, the selection of parents in mutation has been verified to be critical for the DE performance. In this paper, we have studied the differential evolution algorithm from the point of the view of mutation operators and have been proposed two mutation operators to improve the performance of the differential evolution algorithm. The main contributions of this dissertation are as follows.(1) In terms of the weak local searching ability of the differential evolution algorithm as well as the excellent individuals in the population are not effectively used to guide the search, a differential evolution mutation operator based on individual sorting and individual proximity is proposed. In the mutation operator, the higher ranking a parent obtains, the more opportunity it will be selected as base vector. The parent which is closer to the base vector, the more opportunity it will be selected as the terminal of the difference vector. Randomly select the starting point of the difference vector in the population. The proposed operator is applied to the original DE and advanced DE, and the standard test function test results show that the proposed algorithm can effectively improve the convergence ability of the corresponding algorithm, and improve the performance of the algorithm.(2) In terms of the parents in the mutation operators are randomly chosen from the current population generally does not take full advantage of the current population of outstanding individual information, and the search space of important information, a differential evolution mutation operator based on individual sorting and individual grouping is proposed. The population of all individuals can be sorted in ascending order from the best to worst based on its fitness. The sorted population is divided into three groups. The parents of the mutation operator are selected in the different groups, so as to extract the maximum search space information. The base vector is selected in the first fittest group, the higher ranking a parent obtains in the first group, the more opportunity it will be selected as base vector. The terminal of the difference vector is selected in the second best group, the higher ranking a parent obtains in the second group, the more opportunity it will be selected as terminal of the difference vector. The starting point of the difference vector is selected randomly in the remaining set. The proposed mutation operator not only extracts the information of the search space, but also uses the excellent individual information of the population. It is a good balance of the overall exploration ability and the ability of exploitation. The proposed operator is applied to the original DE and advanced DE, the standard test function test results show that the proposed algorithm can effectively improve the performance of the corresponding algorithm.