Adaptive Differential Evolution with Directional Information Based Search Moves

Differential Evolution (DE) is one of the most simple and efficient Evolutionary Algorithms exist till now for global optimization problems. It has reported exceptionally good results when tested over all the benchmark problems and some of the real world problems, although it suffers from the troubles of slow and premature convergence. Generally the performance of DE is sensitive to the choice of mutation and crossover strategies and their associated control parameters. In this paper we propose a DE called Adaptive Differential Evolution with Directional Information based Search Moves (ADE-DISM) in which we basically have improved the mutation and crossover strategies adopted in 'DE/rand/1/bin'. In ADE-DISM we varied the control parameters F and CR in an adaptive manner and have introduced a new parameter w. We have used some directional information based moves over the population and introduced a Mean_Best_Vector for mutation purpose. However, the proposed scheme is shown to be statistically significantly better than or at least comparable to several existing DE variants when tested over the CEC 2005 benchmark problems for 30 and 50 dimensions of the problems.