Decomposition Based Differentiate Evolution Algorithm with Niching Strategy for Multimodal Multi-objective Optimization

Comparing to the multi-objective optimization, multimodal multi-objective optimization brings greater challenge since it involves both the decision space and objective space. The classic multi-objective optimization method MOEA/D could not locate more than one optimal solution in the decision space corresponding to the same optimal solution in PF due to the shortage of strategies for handing multimodality. Therefore, a modified MOEA/D whth niching strategy is proposed, in which the MOEA/D-DE is modified for balancing the convergence and diversity in the objective space while the niching strategy is adopted for save the multiple solutions in the decision space. Further more, the redundant deletion strategy works for removing the redundancy and saving the computational resource. The proposed algorithm is tested on the 22 newly proposed benchmark functions. Experimental results show the competitive performance of the proposed algorithm.