Self-adapting differential evolution algorithm with extension variable dimension

Recently, the differential evolution (DE) algorithm has attracted much attention as an effective approach for numerical optimization. Since the performance of DE is sensitive to the choice of associated control parameters, a large number of strategies on parameter determination have been presented in the past several years. However, most of them have limitations. Thus, to get optimal performance, time-consuming parameter tuning is necessary. This paper introduces an extension variable dimension of DE (EVSDE). In EVSDE, the control parameters are considered as a variable of the component. The variable dimension is expended and a new mutation strategy is employed for the extension dimension of variables on mutation operation. On the basis of experience value, the control parameters follow the individual variable and implement the dynamic self-adaptive process in the evolutionary process. It thus helps to improve the robustness of the algorithm and avoid premature convergence. Simulation results show the EVSDE is better than or at least comparable to other classic and adaptive DE algorithms from the literature in terms of convergence performance for a set of 10 benchmark problems.