Microscale Searching Algorithm for Solving MultiModal MultiObjective Optimization Based on Local Pareto Front Detection

Multimodal multiobjective optimization problems with different solutions corresponding to the same objective vector are very common in the real world. Most multimodal multiobjective evolutionary algorithms only focus on solving the global Pareto set and ignore the equally meaningful local Pareto set, while the performance of algorithms that can solve the local Pareto set in the objective space needs to be improved. In this paper, a microscale searing algorithm is proposed for multimodal multiobjective optimization based on local Pareto front detection. The algorithm detects the local Pareto front through a neural network to adaptively adjust the selection strategy and search range, forming a microscale searing algorithm for multimodal multiobjective evolutionary algorithm (MMOEA_MS) framework. In order to verify the effectiveness of MMOEA_MS, we solved 22 problem sets and compared them with the solution results of other algorithms. The experimental results show that MMOEA_MS can more accurately determine whether the current problem has a local Pareto front, and its performance in the objective space is better than other existing algorithms.