高速铁路客运网络收益管理优化模型研究

发布时间：2018-04-05 21:25

本文选题：铁路旅客运输　切入点：收益管理　出处：《清华大学》2016年博士论文

【摘要】：收益管理已经成为现代管理科学与运筹学的一个重要分支,在我国高速铁路基础设施大规模建设的背景下,“四纵四横”高速铁路网络初具规模。然而现有的高速铁路运营管理方法没有根据高速铁路实际运营情况和特点进行调整与改进,制约着高速铁路服务水平的提升。收益管理方法的应用对于提高铁路客运的运营效益、改善对旅客的服务质量等方面具有重要意义。本文通过对大量文献进行整理和总结,回顾了客运专线开行方案和收益管理问题及相关领域的理论研究进展和实际应用状况。在此基础上,提出了一个大规模客运专线网络运营优化模型,并构建了求解此模型的列生成算法和启发式快速算法。目的是将客运专线网路的开行方案优化与动态收益优化问题结合起来,解决更大、更复杂的客运网络运营优化问题。模型以列车运营总收益最大化为目标。用随机生成数据进行的实验表明,模型及算法可以在较短的时间内求解较大规模的运营优化问题。目前我国的高铁车票均是全价票,并没有采用打折售票的模式,但是随着我国高铁网络运力逐步增加,我国的高铁售票系统采用打折售票的模式是必然趋势。为研究票价打折与席位控制问题,本文对上述模型进行了扩展,构建了可以考虑多席位等级、多折扣等级、不同票价等级之间和不同时间段之间客流转移的客运专线网络收益管理优化模型。通过模拟实验分析,可以找到最合理的打折策略。模型还可以同时给出每次列车为不同票价需求预留的席位数,可以作为客票席位控制的参考数据。本文基于FICO XPRESS优化平台对模型和算法进行了实现,利用随机生成的模拟数据进行了数据实验,验证了本论文构建的客运专线网络收益管理优化模型的逻辑正确性,可有效求解大规模客运专线网络的收益管理优化问题。从实验结果看,本论文设计的列生成算法和启发式算法都能在不显著降低求解精度的前提下,减小模型规模,降低求解难度,可以满足实际客运系统的优化计算的需求。
[Abstract]:Revenue management has become an important branch of modern management science and operational research. Under the background of large-scale construction of high-speed railway infrastructure in China, the "four vertical and four horizontal" high-speed railway network is beginning to take shape.However, the existing high-speed railway operation management methods have not been adjusted and improved according to the actual operation conditions and characteristics of high-speed railway, which restricts the improvement of high-speed railway service level.The application of revenue management is of great significance in improving the efficiency of railway passenger transportation and the service quality of passengers.In this paper, a large number of literatures are summarized, and the development of theoretical research and practical application in related fields of passenger dedicated line operation scheme and revenue management are reviewed in this paper.On this basis, a large scale passenger dedicated line network operation optimization model is proposed, and a column generation algorithm and a heuristic fast algorithm for solving this model are constructed.The purpose of this paper is to combine the optimization of passenger dedicated line network with the optimization of dynamic income to solve the larger and more complex problem of passenger network operation optimization.The model aims at maximizing the total revenue of train operation.Experiments with randomly generated data show that the model and algorithm can solve large scale operation optimization problems in a short time.At present, all the high-speed railway tickets in our country are full price tickets, but with the increasing capacity of high-speed rail network in our country, it is an inevitable trend for the high-speed railway ticket system in our country to adopt the mode of discount ticketing.In order to study the problem of discount and seat control, this paper extends the above model and constructs a multi-seat class and multi-discount class.The optimal model of revenue management of passenger dedicated line network between different fare levels and different time periods.Through the simulation experiment analysis, we can find the most reasonable discount strategy.The model can also give the number of seats reserved for each train for different fare requirements, which can be used as the reference data for ticket seat control.In this paper, the model and algorithm are implemented based on FICO XPRESS optimization platform, and the simulation data generated by random are used to carry out the data experiment, which verifies the logic correctness of the revenue management optimization model of passenger dedicated line network constructed in this paper.It can effectively solve the problem of revenue management optimization of large scale passenger dedicated line network.From the experimental results, both the column generation algorithm and the heuristic algorithm designed in this paper can reduce the scale of the model and the difficulty of the solution without significantly reducing the accuracy of the solution, and can meet the needs of the optimization calculation of the actual passenger transportation system.
【学位授予单位】：清华大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：F532

【相似文献】