期刊
COMPUTERS & OPERATIONS RESEARCH
卷 80, 期 -, 页码 159-172出版社
PERGAMON-ELSEVIER SCIENCE LTD
DOI: 10.1016/j.cor.2016.11.020
关键词
Airline crew pairing; Column generation; Branching methods
类别
资金
- CRSNG
- AD OPT (Division of Kronos) [RDCPJ 477127-14]
Airline crew scheduling is typically performed in two steps : crew pairing followed by crew assignment. The crew pairing problem (CPP) finds a set of pairings (sequences of flights separated by connections or rests starting and ending at the same crew base) that covers a set of flights at minimum cost. The crew assignment problem consists of assigning the crew members to these pairings to create their individual schedules. The main downside of this sequential approach is that the pairings generated in the first step are not all suitable for the crew assignment step, yielding poor-quality solutions. This paper studies an extension of the CPP that includes additional constraints limiting the total worked time at each crew base. This problem, called the CPP with base constraints (CPPBC), is designed to improve the coupling of the two scheduling steps. To solve the CPPBC, we develop four branch-and-price heuristics: three of them rely on known heuristic branching schemes, the other introduces a new branching method, called retrospective branching. This branching scheme is designed to detect and revise poor branching decisions made earlier in the search tree, without backtracking. We tested and compared these four heuristics on real-world datasets. Our results show that the algorithm with retrospective branching yields, most of the times, better-quality solutions than the other tested methods. (C) 2016 Elsevier Ltd. All rights reserved.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据