A Hybrid Multiobjective Differential Evolution Algorithm Based on Improved e-Dominance

Differential Evolution(DE) is a kind of simple but powerful evolutionary optimization algorithm with many successful applications. However, it has some weaknesses, especially the slow convergence speed because of weak local search ability in its stochastic search. To overcome the drawback, we first employ the orthogonal design method with quantization technique to generate the initial population, and then incorporate descent direction search of traditional optimization method into DE algorithm to improve the ability of DE in the process of solving multiobjective optimization problems(MOPs), where the descent direction can be found by using the dominance relationship among individuals. On the other hand, to obtain uniformly spread nondominated solutions and avoid deleting the extreme points, an improved $\epsilon$-dominance strategy is proposed to update the external nondominated archive. Finally, experiment results confirm the effectiveness of the proposed algorithm.