基于字典排序和解关联的多目标优化算法

Multi-objective Optimization Algorithm Based on Lexicographic Sorting and Solution Association

基于分解的多目标进化算法(MOEA/D)正成为一种越来越受欢迎的多目标优化算法.但是它也有一些不足之处.例如,每一个新生成的解将会对多个相邻的子问题中的解进行更新,这样同一个解可能关联好几个子问题,会引起多样性的损失.因此,提出了一种MOEA/D的改进算法,先通过字典排序选出子代种群,以保证种群的多样性.然后,通过一种基于解与权向量之间角度信息的关联过程将解与子问题相关联,以此来提高算法的收敛效率.所提出的算法与其它一些多目标优化算法在2至3目标的基准测试问题上进行了比较.实验表明,所提出的算法优于与之比较的其它算法.

Multi-objective evolutionary algorithm based on decomposition （ MOEA/D）, has become an increasingly popular framework for evolutionary multi-objective optimization, but it has some disadvantages to overcome. For example, each newly generated solution will update the solutions from the neighboring subproblems. So a solution may simultaneously associate with multiple solutions under such circumstances and the diversity loss in the population is unavoidable. In this paper, solutions are selected based on a lexicographic sorting to ensure the diversity in the population, and each selected solution is associated with a subproblem based on the angle between its objective vector and the subproblem＇s weight vector to improve the convergence efficiency of the proposed algorithm. The proposed algorithm is compared with some other multi-objective optimization algorithms on the benchmark problems with two or three objec- fives. The experimental results show that the proposed algorithm outperforms others.