Differential evolution algorithm with survival of fitness strategy

In order to overcome the inferiority that the Differential Evolution algorithm is easy to fall into the local optimum, an improved Differential Evolution algorithm is proposed. On the basis of Differential Evolution, an eliminating mechanism of biological competition is introduced; on the basis of differential evolution, an elimination mechanism of biological competition is introduced. The framework consists of two categories: The survival of the fitness strategy and the population grading framework with gray wolves optimizer. The survival of the fitness strategy improves the situation of falling into local optimization. This mechanism eliminates m individuals by comparing the fitness values of individual populations after evolutionary variation and randomly generates individuals with the same number of eliminated individuals. In this paper, using the survival of the fitness strategy makes those individuals who fall into local optimization and are at a lower level in the process of exploration no longer participate in the exploration and withdraw from their life cycle in advance. In the population renewal stage, the individuals that are also eliminated no longer participate. The CEC2017 competition on single objective bound-constrained numerical optimization and four state-of-the-art DE variants are employed to investigate the effectiveness of the proposed algorithm. Experimental results show that Differential Evolution with Survival of Fitness Strategy (SFDE) is competitive in terms of accuracy and convergence.

I. Introduction

Differential evolution algorithm (DE) was first proposed by Storn and Price [1]. DE stands out among many algorithms with its simple structure and high performance. At the first IEEE International Conference on Evolutionary Computation, which has concerned many scholars because of its excellent performance. However, DE algorithms mainly focus on exploitation capability, so they may not be effective in solving different types of problems.