RDE: Improving Differential Evolution by Using Ranking Information of Search Points

Tetsuyuki TAKAHAMA  Setsuko SAKAI  Akira HARA 

Publication
D - Abstracts of IEICE TRANSACTIONS on Information and Systems (Japanese Edition)  Vol.J95-D  No.5  pp.1196-1205
Publication Date: 2012/05/01
Online ISSN: 1881-0225
Print ISSN: 1880-4535
Type of Manuscript: PAPER
Category: 
Keyword: 
Differential Evolution,  control parameters,  evolutionary computation,  nonlinear optimization, 

Full Text(in Japanese): PDF(558.3KB)

Summary: 
Differential Evolution (DE) is an efficient and robust evolutionary algorithm and has been successfully used in various application fields. There exist many studies on selecting values for control parameters of DE. In this study, a new method of setting the parameter values of DE by using the ranking information of search points is proposed. In the method, different parameter values are used for each search point from the viewpoint of the convergence and the diversity of search points, although the same values are used for all search points in general. The advantage of the proposed method is shown by solving some benchmark problems and comparing the method with a standard DE, the improved DE (REAL) which modifies the generational model, the efficient DE (potentialDE) which utilizes an approximation model with low accuracy, JADE which tunes the control parameters based on success and SavingMGG which is a real-coded genetic algorithm with an improved generational model.