ESS: Dynamic and Stochastic Network Flow Models for Robust Revenue Optimization in Hotel Service Sector

预期该项目的研究将在方法上作出重大贡献，通过控制旅馆和其他服务行业的预订来加强收入管理做法。特别是，为了增加收入潜力，本项目调查了弥合理想收入解决方案（如果所有请求都提前知道）和当前最大化预期收入的实践之间的收入差距的可能性。为了增强需求假设的现实性，将提出一个用于对客户预订决策（消费、取消和持续时间）建模的统计框架。该项目还将通过确定几种需求需求场景的理想收入解决方案来开发强大的初始解决方案。此外，随着当前预订请求的显示，这个最优先验解决方案将随着时间的推移而更新，从而考虑到预测需求和实际需求之间的差异。通过综合网络、可靠性、算法、模拟和预测等跨学科原则，为酒店业的预订控制开发新的模型和工具，研究结果和模型最终将导致在各种服务行业网络中显著改进和更稳健的收入优化实践的发展。对社会各阶层的好处包括：(1)为预订经理提供决策支持工具；(2)提高酒店和航空公司的收入和收入可靠性；(3)卫生保健服务规划；(3)特殊活动（如音乐会）的预订控制。在航空公司、旅馆和租车公司等服务行业，简单的启发式和经验工具已被应用于预订控制。目前这些模型的假设过于严格，没有考虑到收入管理问题的复杂性。本项目旨在通过基于网络流的方法，开发预订问题的现实建模方法，从而增加收入潜力和鲁棒性。为了实现这些目标，本项目旨在：(1)开发一个分解框架来模拟随时间变化的预订需求；(2)提出动态网络模型，求解实时预约控制问题的鲁棒最优解；(3)开发基于网络的模型，以最大限度地提高不确定电弧成本（由于取消，未显示等）下收入的鲁棒性。将开设研究生水平的“动态和随机基础设施和服务网络的可靠性和优化”课程，为各种服务和基础设施行业的复杂物理和/或虚拟网络的研究和实践做好准备。该项目还将积极招收女性和少数族裔学生，以提高服务行业工程师的多样性。

The research from this project is expected to produce significant methodological contributions to enhance revenue management practices through reservations control, in hotels and other service industries. In particular, to increase revenue potential, this project investigates the possibility of bridging the revenue gap between ideal revenue solution (if all requests were known ahead of time) and the current practice of maximizing expected revenue. To enhance the realism of demand assumptions, a statistical framework for modeling customer reservation decisions (consumption, cancellation, and duration) will be proposed. The project will also develop robust initial solutions by determining ideal revenue solutions for several demand request scenarios. Further, this optimal apriori solution will be updated over time as current reservation requests are revealed, thus accounting for discrepancies between forecasted and actual demands. The research findings and models will ultimately lead to the development of significantly improved and more robust revenue optimization practices in a variety of service industry networks, by synthesizing interdisciplinary principles from networks, reliability, algorithms, simulation, and forecasting to develop new models and tools for reservations control in the hotel industry. Benefits to various segments of the society include: (1) decision support tools for reservations managers; (2) improved revenues and revenue reliability for hotels and airlines; (3) health care service planning; and (3) reservations control for special events (e.g. concerts). In service sector industries such as airlines, hotels and rental car agencies, simple heuristic and empirical tools have been applied for reservations control. These current models are too restrictive in their assumptions, and do not account for the complexity of the revenue management problem. This project aims to develop methods for realistic modeling of the reservations problem, thus leading to increased revenue potential and robustness, through a network flow-based approach. To achieve these objectives, this project aims to: (1) develop a disaggregate framework to model the demand for reservations over time; (2) propose dynamic network models to obtain robust optimal solutions to the real-time reservations control problem; and (3) develop network-based models to maximize the robustness of revenues under uncertain arc costs (due to cancellation, no shows etc.). A graduate level course on 'Reliability and Optimization of Dynamic and Stochastic Infrastructure and Service Networks' will be developed to prepare leaders in research and practice in various service and infrastructure industries operating complex physical and/or virtual networks. The project will also aggressively recruit female and minority students to enhance the diversity of engineers in the service sector.