返回
A beam search algorithm for the biobjective container loading problem
DOI:10.1016/j.ejor.2020.03.040.png)
摘要
En 中文
In this work we deal with a biobjective variant of the single container loading problem. The problem consists in packing a set of boxes into a large box or container. In addition to the usual objective related to maximize the volume utilization, we include a second objective consisting on maximizing the total profit (e.g., weight or priority) of the loaded boxes. We propose to adapt bsg, a state-of-the-art single-objective algorithm for solving the biobjective problem. bsg can be seen as an incomplete version of a breadth first search in which each level of the search tree is restricted to a fixed number of promising nodes. In order to select the most promising nodes, the algorithm constructs a solution from them by using a greedy algorithm and evaluates the objectives. To deal with two objectives, we propose to select nodes by using well-known multi-objective criteria based on dominance and crowding distance related to the greedy evaluations. We go a little bit further and orient the greedy algorithm to one objective or the other by using a mechanism which changes the configuration of the heuristic function dynamically. The proposal is compared with bsg in a set of well-known instances reporting a 24% increase in the hypervolume indicator. Compared to other biobjective approaches, the reported results are also promising. Codes and test instances are available in: https://github.com/rilianx/Metasolver#bo-bsg (C) 2020 Elsevier B.V. All rights reserved.
Keyword:
multiobjective optimization
container loading problem
beam search
greedy
AI总结
对已上传原文的论文进行重点信息的提取,主要内容包括:简要概述、研究摘要、背景介绍、关键亮点、图文解析、展望与总结。
期刊
IF:
6
论文数:
2.2W
被引数:
6.4W
机构
引用论文
The six elements to block-building approaches for the single container loading problem
APPLIED INTELLIGENCE
IF3.5

