A multi-objective bilevel optimisation evolutionary algorithm with dual populations lower-level search
Weizhong Wang,Hai-Lin Liu,Hongjian Shi
DOI: https://doi.org/10.1080/09540091.2022.2077312
2022-06-03
Connection Science
Abstract:In multi-objective bilevel optimisation problems, the upper-level performance of different lower-level optimal solutions may be very different, even though they belong to the same lower-level problem. It may lead to poor optimisation results. Therefore, the lower-level search should search lower-level non-dominated solutions that are also non-dominated in the upper-level objective space. In this paper, we use two populations in the lower-level search. The first population maintains non-dominance and diversity in the lower-level objective space and provides the second population with convergence pressure from the lower level. The second population selects the upper-level non-dominated solutions that are not dominated by the first population in the lower-level objective space, which make the second population maintain the non-dominance at both upper and lower levels. Besides, to improve the search efficiency, we set up the upper-level mating pool to generate the upper-level vectors of offsprings near the upper-level vectors of the better individuals in the current population. To balance convergence and diversity, the selection operator of a decomposition based multi-objective evolutionary algorithm is adopted. The proposed algorithm has been evaluated on a set of benchmark problems and a real-world optimisation problem. Experimental results demonstrate that the proposed algorithm is efficient and effective.
computer science, artificial intelligence, theory & methods