Abstract:
In order to solve the problem of how to balance convergence and diversity of MOEAs, an adaptive mutation multi-objective optimization strategy based on diverse initial population (IM) is proposed. This strategy can be combined with MOEAs based on Pareto dominance and decomposition. IM uses the method of diversification of initial population, which makes the initial population of the algorithm have better diversity. Then, in the process of population evolution, adaptive mutation method is used to reduce the mutation probability of excellent individuals, increase the mutation probability of poor individuals and improve the convergence rate of population. In the experiment part, the MOEA/D-DE, NSGA-Ⅲ and NSGA-Ⅱ algorithms of IM strategy are compared with the original MOEA/D-DE, NSGA-Ⅲ and NSGA-Ⅱ algorithms, and IM- MOEA/D-DE and IM-NSGA-Ⅲ are taken as examples to compare with the other four typical multi-objective optimization algorithms MOEA/D-Pas、MOEA/DD、SPEA/R and VaEA.The results show that IM strategy can improve the convergence and diversity of MOEA/D-DE, NSGA-Ⅲ and NSGA-Ⅱ algorithms, and it has a strong competitiveness compared with the existing classical multi-objective algorithm.
