工业工程

工业工程 ›› 2024, Vol. 27 ›› Issue (5): 1-10.doi: 10.3969/j.issn.1007-7375.230171

• 服务运营管理与调度优化 •    

考虑旅客选择行为的收益管理和机型指派联合模型

乐美龙1, 陈怡1, 黄周春2   

  1. 1. 南京航空航天大学 1. 民航学院;
    2. 经济与管理学院，江苏 南京 211106
  • 收稿日期:2023-09-11 发布日期:2024-11-05
  • 通讯作者: 黄周春 (1986—)，男，安徽省人，副教授，博士，主要研究方向为随机规划、组合优化和大规模整数规划及其在能源系统优化和航空运营管理领域的应用。Email: zhouchun.huang@nuaa.edu.cn E-mail:zhouchun.huang@nuaa.edu.cn
  • 作者简介:乐美龙 (1964—)，男，浙江省人，教授，博士，主要研究方向为航空运输系统规划与运营控制优化。Email: meilongle@hotmail.com
  • 基金资助:
    教育部人文社会科学研究青年基金资助项目 (22YJC630041)；南京航空航天大学研究生创新基地开放基金资助项目 (xcxjh20220721)

A Joint Model of Revenue Management and Aircraft Assignment Considering Passenger Choice Behavior

LE Meilong1, CHEN Yi1, HUANG Zhouchun2   

  1. 1. College of Civil Aviation;
    2. College of Economics and Management, Nanjing University of Aeronautics and Astronautics, Nanjing 211106, China
  • Received:2023-09-11 Published:2024-11-05

PDF

12

摘要： 传统收益管理假设旅客需求是独立的，未能充分考虑到旅客的选择行为，本文讨论了考虑旅客选择行为的收益管理问题和机型指派联合优化问题。根据对历史销售数据的观察，发现旅客对最低价舱位具有特殊偏好，因此本文利用尖峰多项Logit (spiked multinomial logit, Spiked-MNL) 选择模型来预测旅客潜在真实需求，并以航空公司利润最大化为目标，构建考虑旅客选择行为的收益管理和机型指派联合模型。本研究利用某航空公司的实际销售交易数据进行算例分析，验证所提出的模型的可行性，并通过在不同场景下的比较来评估模型的性能。结果显示，在比较联合模型和独立模型的利润时，联合模型的平均利润高出5%；在不同机队规模的比较中，5机型联合模型产生的利润最高，其平均利润比3机型高出6.6%，比9机型高出7.3%。此外，与以往关于利用旅客选择模型预测旅客需求的方法相比，Spiked-MNL模型能够更准确地反映旅客的实际购买行为，其平均利润比广义吸引力模型 (generalized attraction model, GAM) 高出0.7%，比多项Logit (multinomial logit, MNL) 模型高出1.2%。

关键词: 网络收益管理, 机型指派, 旅客选择行为, 联合优化

Abstract: The assumption of traditional revenue management on passenger demand is independent and fails to fully account for passenger choice behavior. This paper discusses the revenue management and aircraft assignment problems considering passenger choice behavior. Observations from historical sales data reveal that passengers have a special preference for the lowest fare class. Therefore, the Spiked Multinomial Logit (Spiked -MNL) choice model is used in this paper to predict the potential real demand of passengers. Also, a joint model of revenue management and aircraft assignment considering passenger choice behavior is established, with the objective of maximizing airline profit. Case studies are conducted using actual sales transaction data from an airline to verify the feasibility of the proposed model and to evaluate its performance by comparison in different scenarios. Results show that the joint model yields an average profit of 5% higher than the independent model. In the comparison of different fleet sizes, the 5-aircraft joint model produces the highest profit, with an average profit of 6.6% higher than the 3-aircraft model and 7.3% higher than the 9-aircraft model. In addition, compared with traditional methods of predicting passenger demand using passenger choice models, the Spiked-MNL model can reflect actual passenger purchasing behavior more accurately, with an average profit of 0.7% higher than the Generalized Attraction Model (GAM) and 1.2% higher than the Multinomial Logit (MNL) model.

Key words: network revenue management, aircraft assignment, passenger choice behavior, joint optimization

中图分类号: