Dynamic Multi-modal Multi-objective Evolutionary Optimization Algorithm Based on Decomposition.

In this paper, the independent convergent and non-convergent decision variables are firstly obtained by analyzing the contribution of decision variables to the objective function based on the existing research results of multi-objective optimization algorithms. Secondly, according to their characteristics, the multi-population is employed, so that the population can search the corresponding multiple Pareto optimal solution set in each individual environment. Then, when the problem changes, two more targeted response strategies are proposed for different types of decision variables and their effects on the objective function. As the environment changes, the algorithm can ensure the rapid convergence of the population in the objective space, while maintaining the diversity of the population in the decision space and the objective space. Therefore, the proposed algorithm has the ability of quickly respond to the change of the problem and maintain the diversity of the solution set.