A decomposition multi-objective evolutionary algorithm based on vector arrangement and strategy adaptation mechanism

When a decomposition-based multi-objective evolutionary algorithm is used to solve complex multi-objective optimization problems (MOPs), the distribution of weight vectors cannot adapt to the different shapes of Pareto front surfaces, resulting in a poor distribution of feasible solutions. This paper proposes a decomposition multi-objective evolutionary algorithm (MOEA/D-VAS) based on vector programming and a strategy adaptive mechanism (SaM). In this algorithm, virtual vectors are added to the initial decomposition vector set, and weight vectors are screened by Pareto advantage and an adaptive function to improve the diversity of solutions. Additionally, the differential evolution operator uses SaM to select different greedy strategies for each individual, which improves the convergence speed of the algorithm and the accuracy of the weight vector to predict the true Pareto surface while effectively balancing the diversity and convergence of the solution. DTLZ and WFG are used to compare different algorithms. The results show that the convergence and distribution of the proposed algorithm are greatly improved, and the MOEA/D-VAS algorithm can adapt to more types of MOPs.