Differential evolution with neighborhood-based adaptive evolution mechanism for numerical optimization

This paper presents a novel differential evolution algorithm for numerical optimization by designing the neighborhood-based mutation strategy and adaptive evolution mechanism. In the proposed strategy, two novel neighborhood-based mutation operators and an individual-based selection probability are developed to adjust the search performance of each individual suitably. Meanwhile, the evolutionary dilemmas of the neighborhood are identified by tracking its performance and diversity, and alleviated by designing a dynamic neighborhood model and two exchanging operations in the proposed mechanism. Furthermore, the population size is adaptively adjusted by a simple reduction method. Differing from differential evolution variants based on neighborhood and evolutionary state, the proposed algorithm makes full use of the characteristics of individuals, identifies and alleviates the neighborhood evolutionary dilemmas of each individual. Compared with 21 typical algorithms, the numerical results on 30 benchmark functions from CEC2014 show that the proposed algorithm is reliable and has better performance. (C) 2018 Elsevier Inc. All rights reserved.