A Simple Differential Evolution with Random Mutation and Crossover Constants for Constrained Optimization

Abstract

The article proposes a simple version of the differential evolution algorithm (abbreviated as sDE) in which the mutation factor and crossover constant are chosen randomly in the range (0,1) during the search for the optimal solution. The sDE is the same as the original version of the differential evolution algorithm, except the user does not have to choose the best values of mutation constant and crossover constant for each optimization problem. Therefore, the optimization process is now very simple as it remains only one parameter (i.e. the population size) in the algorithm, besides the stopping criterion (e.g. number of iterations). It also consumes less computation time than the original differential evolution as it is not necessary to tune the mutation and crossover constants. In this study, the proposed technique is applied to three constrained optimizations, three engineering design problems, and six planar and spatial trusses under frequency constraints. Despite the very simple characteristics of the proposed technique, sDE gives promising results in comparison with other results in the literature.