Improved Differential Evolution with Searching Pioneer for Solving Multi-modal Optimization Problems

The first thing to apply differential evolution (DE) for solving optimization problems is to decide several parameters. These parameters' setting will influence its solution searching ability directly. Thus, how to decide suitable parameters for current solution searching stage is an issue for DE. In this study, the searching pioneer is proposed. The vector which performed best performance in previous iteration will share its parameters information to other vectors for guiding them toward to potential solution space. Also, the elitist crossover is also involved to combine useful information to generate potential vectors. In order to compare the proposed method with other DE variants, in experiments, five multi-modal test functions of CEC 2005 are adopted. The experiments showing clear proof that the proposed method performs better than other related works on most test functions.