Adaptive CR values with fuzzy inference system for differential evolution algorithm

This study proposed fuzzy-based adaptive method for crossover rate (CR) values to improve the convergence rate of Differential Evolution (DE) algorithms. The proposed method decides the CR values according to two factors of the evolutionary environment, the change rate of solution of center of solutions and the density of each dimension. The proposed DE algorithm is tested by the CEC2015 benchmark suits and compared with fuzzy adaptive differential evolution (FADE). The experimental results showed that the proposed method really improved performance for 11 of 15 evaluated functions in 10, 30, and 50 dimensions and enhanced the convergence.