Differential Evolution Using Interpolation Strategy for Multimodal Multiobjective Optimization

The ultimate goal of solving Multimodal Multiobjective Optimization Problems (MMOPs) is to identify all Pareto optimal solution sets that are equivalent. Strong search capability and environmental selection mechanism must be possessed by an algorithm designed to solve MMOPs. In this paper, we propose a novel algorithm called Differential Evolution using Interpolation Strategy for Multimodal Multiobjective Optimization (MMODE_IS). The proposed algorithm comprises three main components. First, a neighborhood search strategy that using the data interpolation technique is proposed to aid population evolution and enhance the algorithm's search capability in the decision space. Second, the algorithm dynamically chooses the environment selection mechanism based on the population's evolutionary stage through a novel adaptive selection strategy, thus selecting outstanding individuals that are beneficial to the population's evolution. Finally, we evaluated the performance of MMODE_IS by comparing it with several state-of-the-art algorithms. The experimental results prove that MMODE_IS can effectively solve the MMOPs.