考虑旅客选择行为的高铁席位动态控制策略
|
摘要
利用基于选择行为的网络收益管理原理研究高铁席位动态控制策略.首先,针对多列车多停站的高铁网络建立考虑旅客选择行为的席位控制动态规划模型.然后,设计两阶段控制机制:第一,采用近似动态规划技术,对价值函数进行线性近似,设计策略迭代算法离线获得时间依赖的投标价格;第二,采用能够描述多种选择行为机理的马尔科夫链选择模型刻画旅客选择行为,将投标价格作为输入参数,在线运行品类优化算法获得实时控制策略.最后以京沪高铁为背景设计仿真实验,结果表明该机制在需求水平较低时能够明显改善收益,且保护长途票额.该机制不仅可获得实时动态控制策略,而且通用性好,还可利用历史数据进行参数学习与更新,实现更为精细地连续优化.
Dynamic seat inventory control strategy for high-speed rail passenger transportation is studied with choice-based network revenue management theory. Firstly, the dynamic programming model of seat inventory control for high-speed rail network with multi-trains and multi-stops is established. Secondly, two-stage seat inventory control mechanism is designed. At the offline stage, the value function is approximated linearly with approximate dynamic programming technique, and the time-dependent bid price is obtained with the policy iterative algorithm. At the online stage, the bid price is taken as the input parameter, and the real time control strategy is obtained by the assortment optimization algorithm of the Markov Chain Choice Model. Finally,simulation experiments with different parameters were carried out under the background of Beijing-Shanghai highspeed railway. The results show that the strategy optimization mechanism can improve the profit significantly and protect long distance tickets when the demand level is low. This mechanism not only provide dynamic seat control stagey with better generality, but also can use historical data to get learning bid price and choice parameters and achieve continuous dynamic optimization.
Dynamic seat inventory control strategy for high-speed rail passenger transportation is studied with choice-based network revenue management theory. Firstly, the dynamic programming model of seat inventory control for high-speed rail network with multi-trains and multi-stops is established. Secondly, two-stage seat inventory control mechanism is designed. At the offline stage, the value function is approximated linearly with approximate dynamic programming technique, and the time-dependent bid price is obtained with the policy iterative algorithm. At the online stage, the bid price is taken as the input parameter, and the real time control strategy is obtained by the assortment optimization algorithm of the Markov Chain Choice Model. Finally,simulation experiments with different parameters were carried out under the background of Beijing-Shanghai highspeed railway. The results show that the strategy optimization mechanism can improve the profit significantly and protect long distance tickets when the demand level is low. This mechanism not only provide dynamic seat control stagey with better generality, but also can use historical data to get learning bid price and choice parameters and achieve continuous dynamic optimization.
引文
[1]TALLURI K T,VAN RYZIN G J.Revenue management under a general discrete choice model of consumer behavior[J].Management Science,2004,50(1):15-33.
[2]乔珂,赵鹏,文佳星.基于潜在类别模型的高铁旅客市场细分[J].交通运输系统工程与信息,2017,17(2):28-34.[QIAO K,ZHAO P,WEN J X.Passenger market segmentation of high-speed railway based on latent class model[J].Journal of Transportation Systems Engineering and Information Technology,2017,17(2):28-34.]
[3]张小强,李煜,黄帅勋,等.基于最大凹向包络的高速铁路客运动态定价策略[J].交通运输系统工程与信息,2016,16(6):1-8.[ZHANG X Q,LI Y,HUANG S X,et al.Maximum concave envelope theory based dynamic pricing for railway passenger transportation[J].Journal of Transportation Systems Engineering and Information Technology,2016,16(6):1-8.
[4]赵翔,赵鹏,李博.多列车多停站方案条件下高速铁路票额分配研究[J].铁道学报,2016,38(11):9-15.[ZHAO X,ZHAO P,LI B.Study on high-speed railway ticket allocation under conditions of multiple trains and multiple train stop plans[J].Journal of the China Railway Society,2016,38(11):9-15.]
[5]LIU Q,VAN RYZIN G.On the choice-based linear programming model for network revenue management[J].Manufacturing and Service Operations Management,2008,10(2):288-310.
[6]KOCH S.Least squares approximate policy iteration for learning bid prices in choice-based revenue management[J].Computers&Operations Research,2017,77(January 2017):240-253.
[7]MIRANDA-BRONT J J,MENDEZ-DIAZ I,VULCANOG A.A column generation algorithm for choice-based network revenue management[J].Operations Research,2007,57(3):769-784.
[8]钱丙益.基于旅客选择行为的客运专线收益管理研究[D].成都:西南交通大学,2014.[QIAN B Y.Research on revenue management for dedicated passenger line based on passenger choice behavior[D].Chengdu:Southwest Jiaotong University,2014.
[9]BLANCHET J,GALLEGO G,GOYAL V.Fourteenth ACM conference on electronic commerce,a markov chain approximation to choice modeling[C].ACM,2013:103-104.
[10]FELDMAN J B,TOPALOGLU H.Revenue management under the markov chain choice model[J].Operations Research,2017,65(5):1322-1342.
[2]乔珂,赵鹏,文佳星.基于潜在类别模型的高铁旅客市场细分[J].交通运输系统工程与信息,2017,17(2):28-34.[QIAO K,ZHAO P,WEN J X.Passenger market segmentation of high-speed railway based on latent class model[J].Journal of Transportation Systems Engineering and Information Technology,2017,17(2):28-34.]
[3]张小强,李煜,黄帅勋,等.基于最大凹向包络的高速铁路客运动态定价策略[J].交通运输系统工程与信息,2016,16(6):1-8.[ZHANG X Q,LI Y,HUANG S X,et al.Maximum concave envelope theory based dynamic pricing for railway passenger transportation[J].Journal of Transportation Systems Engineering and Information Technology,2016,16(6):1-8.
[4]赵翔,赵鹏,李博.多列车多停站方案条件下高速铁路票额分配研究[J].铁道学报,2016,38(11):9-15.[ZHAO X,ZHAO P,LI B.Study on high-speed railway ticket allocation under conditions of multiple trains and multiple train stop plans[J].Journal of the China Railway Society,2016,38(11):9-15.]
[5]LIU Q,VAN RYZIN G.On the choice-based linear programming model for network revenue management[J].Manufacturing and Service Operations Management,2008,10(2):288-310.
[6]KOCH S.Least squares approximate policy iteration for learning bid prices in choice-based revenue management[J].Computers&Operations Research,2017,77(January 2017):240-253.
[7]MIRANDA-BRONT J J,MENDEZ-DIAZ I,VULCANOG A.A column generation algorithm for choice-based network revenue management[J].Operations Research,2007,57(3):769-784.
[8]钱丙益.基于旅客选择行为的客运专线收益管理研究[D].成都:西南交通大学,2014.[QIAN B Y.Research on revenue management for dedicated passenger line based on passenger choice behavior[D].Chengdu:Southwest Jiaotong University,2014.
[9]BLANCHET J,GALLEGO G,GOYAL V.Fourteenth ACM conference on electronic commerce,a markov chain approximation to choice modeling[C].ACM,2013:103-104.
[10]FELDMAN J B,TOPALOGLU H.Revenue management under the markov chain choice model[J].Operations Research,2017,65(5):1322-1342.