Differential Evolution with Control Parameters Selected from the Previous Performance
Yen-Ching Chang · 2018
The control parameters of differential evolution (DE) have a big influence on the optimization performance. In general, the optimal setting of control parameters for one objective function is not suitable for another. Therefore, how to choose these control parameters for different objective functions is a very important issue. In this work, a two-stage optimization strategy is adopted to achieve the best performance as far as possible. In the first stage, a grid (10 × 9) of control parameters consisting of the differential weight and the crossover rate are used to the standard DE and the control parameters with the minimum function value are found out. In the second stage, the optimal setting of control parameters is adopted to the remaining iterations. Experimental results show that the proposed strategy contributes to searching the optimal value at the expense of computational time.