Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

Differential evolution has shown success in solving different optimization problems. However, its performance depends on the control parameters and search operators. Different from existing approaches, in this paper, a new framework which dynamically configures the appropriate choices of operators and parameters is introduced, in which the success of a search operator is linked to the proper combination of control parameters (scaling factor and crossover rate). Also, an adaptation of the population size is adopted. The performance of the proposed algorithm is assessed using a well-known set of constrained problems with the experimental results demonstrating that it is superior to state-of-the-art algorithms