基于旅客选择行为的客运专线收益管理研究

发布时间：2018-08-16 10:37

【摘要】：为满足城际间日益增长的旅客运输需求,我国在建设“四纵四横”客运专线的同时,正在进一步延伸并扩大快速客运网络的覆盖范围。客运专线的建设和运营,很大程度上缓解了铁路客运能力瓶颈问题,也不可避免地加剧了国内客运市场各种运输方式之间的竞争,导致客运专线市场化运营趋势进一步得到体现。美、法、英、德等国家铁路客运公司的多年来市场化运营经验表明,收益管理是优化运力结构和资源配置、提高铁路竞争力和运营收益的有效方法。因此,开展客运专线收益管理研究,尤其是基于旅客选择行为、而不是仅从客运专线公司角度开展收益管理研究,不仅体现了以旅客为本的思想,也隶属于收益管理研究中的热点,具有重要的理论和现实意义,为客运专线实施收益管理提供了借鉴和新视角。本文第一章从理论研究和规划实践两个方面分析了选题背景,阐述了客运专线实施收益管理的可行性、必要性和开展收益管理研究的意义,并在对铁路收益管理应用和研究现状系统分析的基础上阐述了有待研究的问题。第二章对收益管理和旅客选择行为的研究成果进行了综述,从而有效把握了收益管理的发展历程、基本内涵、研究内容及基础模型,明确了动态定价和存量控制两种收益管理手段的研究现状和前沿动态,了解了消费者选择行为基础理论和收益管理中常见的消费者选择行为类型,为后文展开提供了研究依据,并奠定了工作基础。第三章构建了旅客对客运专线产品的总体评价关于安全、舒适、速度、频率、准点、价格、便捷七项属性评价的混合回归模型,并采用EM算法估计回归系数,根据贝叶斯统计理论计算旅客属于各类型的概率,利用贝叶斯信息标准和赤池信息标准判定旅客种类个数。结合武广客运专线旅客问卷调查数据,将旅客市场细分为四种类型,并对不同类型旅客的社会经济特征和出行需求特征进行关联分析。第四章借助保留价格刻画旅客选择行为,并假设预订相同路径客票的旅客的保留价格独立且服从同一分布,结合已知票价集合和需求概率,根据Bellman最优化原理,构建动态规划模型,使得可通过票价的动态调整来最优化客运专线期望收益。进一步证得,客票的最优价格随着其边际期望收益的增大而提高,单区段和两区段客运专线的最优票价策略具有阀值特性。第五章针对两个票价等级、多个区段的客运专线收益优化问题,假设旅客存在buy-up行为且其概率已知,在给定客票价格和需求分布的条件下,以各路径客票的嵌套式预订限制为决策变量,构建带约束的非线性整数规划模型,并通过解生成点的获取、初始粒子群的生成、粒子适应值计算、粒子位置更新这四个步骤,实现了模型的求解。第六章针对多区段、多车次、多票价等级的客运专线的收益优化问题,借助偏好序刻画旅客的选择行为,在给定客票和旅客类别特征的条件下,以每个预售时段客票的控制策略为决策变量,构建动态规划模型。鉴于模型规模较大,构建基于选择的确定性线性规划模型对其进行近似,并采用列生成算法和遗传模拟退火算法求解,进而利用所得最优对偶解对原动态规划模型进行启发式分解,最终实现了客运专线期望收益的近似优化。
[Abstract]:In order to meet the increasing passenger transport demand between cities, China is extending and expanding the coverage of the rapid passenger transport network while building the "four vertical and four horizontal" passenger dedicated lines. Competition among various modes of transport in the airport has led to the further realization of the market-oriented operation trend of passenger dedicated lines. The experience of the United States, France, Britain, Germany and other national railway passenger transport companies over the years has shown that revenue management is an effective way to optimize the structure of transport capacity and resource allocation, and to enhance the competitiveness of the railway and operating income. Research on revenue management of PDL, especially on the basis of passenger choice behavior, rather than just from the perspective of PDL company, not only embodies the idea of passenger-oriented, but also belongs to the hotspot of revenue management research. It has important theoretical and practical significance, and provides a reference for the implementation of revenue management of PDL. The first chapter of this paper analyzes the background of the topic selection from two aspects of theoretical research and planning practice, expounds the feasibility, necessity and significance of carrying out revenue management research on passenger dedicated lines, and expounds the problems to be studied on the basis of systematic analysis of the application and research status of railway revenue management. This paper summarizes the research results of revenue management and passenger choice behavior, thus effectively grasping the development process, basic connotation, research content and basic model of revenue management, clarifying the research status and frontier trends of dynamic pricing and inventory control, and understanding the basic theory and benefits of consumer choice behavior. Chapter 3 establishes a mixed regression model of seven attributes evaluation of passenger dedicated line products, including safety, comfort, speed, frequency, punctuality, price and convenience, and estimates the regression coefficient with EM algorithm. According to Bayesian statistical theory, the probability of different types of passengers is calculated, and the number of types of passengers is determined by Bayesian information standard and Chichi information standard.Combining with the questionnaire survey data of passengers on Wuhan-Guangzhou passenger dedicated line, the passenger market is divided into four types, and the socio-economic characteristics and travel demand characteristics of different types of passengers are related. In Chapter 4, the reserved price is used to characterize the behavior of passenger selection, and the reserved price is assumed to be independent and subject to the same distribution. Combining with the known fare set and demand probability, a dynamic programming model is constructed based on Bellman optimization principle to optimize the passenger dedicated transport system through the dynamic adjustment of the fare. It is further proved that the optimal price of passenger tickets increases with the increase of marginal expected revenue, and the optimal fare strategies for single and two-section passenger dedicated lines have threshold characteristics. Under the condition of given ticket price and demand distribution, a nonlinear integer programming model with constraints is constructed by taking nested reservation constraints of each route ticket as decision variables. The model is solved by four steps: obtaining the solution generating points, generating the initial particle swarm, calculating the particle fitness value and updating the particle position. Six chapters are devoted to the revenue optimization of multi-section, multi-train and multi-fare passenger dedicated lines. With the help of preference order, the choice behavior of passengers is described. Under the condition of given ticket and passenger class characteristics, the dynamic programming model is constructed with the control strategy of each pre-booking period as the decision variable. The selected deterministic linear programming model is approximated and solved by column generation algorithm and genetic simulated annealing algorithm. Then the optimal dual solution is used to heuristically decompose the original dynamic programming model. Finally, the approximate optimization of the expected revenue of passenger dedicated line is realized.
【学位授予单位】：西南交通大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：F532.6;U293

【参考文献】