Clustering-Based Differential Evolution with Composite Trial Vector Generation Strategies and Control Parameters

Abstract

Abstract. This paper proposed a novel method Clustering-Based Differential Evolution with Composite Trial Vector Generation Strategies and Control Parameters (C-CODE). It combines several effective trial vector generation strategies with some suitable control parameter settings for get rid of the singleness of control parameters and trial vector generation strategies of traditional differential evolution algorithm. And in order to more effective use of population information is hereby joined the one-step k-means clustering algorithm, so as to improve the performance of the algorithm. Finally, the validity and superiority of the algorithm are verified by 13 international standard test functions.