This paper describes that in multi-objective evolutionary algorithms,convergence and diversity are generally considered when selecting the better solution from the nondominant solution as the offspring.It measures the convergence of the solution by using the L1 norm between the target vector and the ideal point,using the distance between the solutions to represent diversity,and the relationship between all solution pairs constitutes a new indicator matrix,called the Distance Convergence ValueMatrix.In addition,it introduces the determinant point procedure to diversify the selection of subsets,which shows that the greater the difference between the elements within the subset,the greater the value of the determinant.It compares the proposed algorithm with four state-of-the-art MOEAs in 21 different common test examples.The empirical results show that the proposed algorithm is universal in various types of test cases and outperforms several state-of-the-art MOEAs.
关键词
进化算法/多样性/多目标优化/行列式点过程
Key words
evolutionary algorithm/diversity/many-objective optimization/determinatal point processes
维普
万方数据