The Theory And Methods For Railway Seat Inventory Control

ABSTRACT:With the gradual expansion of network scale and the increase in operating mileage of Chinese high-speed railway, railway passenger transport capacity will be gradually released and the shortage of railway passenger transportation situation will be eased. Production of rail transportation gradually shifts from extensive to intensive. Under these circumstances, how to fully take advantage of the railway passenger capacity, improve train load factors and operating revenue has become one of the important problems faced by railway passenger transportation sectors. Based on the existing studies and combined with the characteristics of the Chinese railway passenger transportation, this research mainly focuses on railway seat inventory control issues, and the main work in this thesis could be summarized as follows.(1) We constructed a general model for train seat inventory control problems. The characteristics of the Chinese railway passenger transportation were first summarized based on analyzing related research of Chinese and foreign studies. After that, a general model was established for train seats inventory control problems, meanwhile, the solution approaches were discussed.(2) This thesis proposed railway passenger transport short-term demand forecasting methods for different research needs. Firstly, for the research of the deterministic single train seats inventory control, ARIMA model was put forward based on historical passenger ticket booking data mining. Secondly, for the research of multi-trains stochastic demand seats inventory controls, a demand forecasting method of individual-oriented based on passenger choice behavior was proposed. Passenger ticket booking data was mined by Logit model, meanwhile, the heterogeneity of passenger choice behavior was analyzed, and the result showed that passengers'time value is around100Yuan per hour for those who take Beijing-Shanghai high-speed trains. Finally, the behavior of passenger ticket booking was analyzed in a ticket pre-sale period, and the result showed that passengers booking process is a non-homogeneous Poisson process. After that, the simulation approach of the process was proposed. The research of demand forecasting is the basis of the proposed seats inventory control methods.(3) We further constituted seat inventory control methods based on three strategies, non-nested, nested, and bid price for a single train under single price. Firstly, based on independent demand, a deterministic multi-stage model and a stochastic multi-stage model were established for a single train under single price based on the non-nested, nested, and bid price strategies. Secondly, based on simulation and the software Lingo12.0, algorithms for different control strategies were given. Finally, a comparison of different control methods was made by the train on traditional lines and high-speed lines, and the applicable conditions of the control strategy were given.(4) Moreover, this paper built a seats inventory control model based on passenger choice behavior for multi-train under single price. Meanwhile, a column generation algorithm was designed to solve the model. Firstly, passenger choice behavior was analyzed for multi-train under a single price. Secondly, a dynamic model of the multi-train seats inventory control was established. Thirdly, with the huge number of variables in the model, a column generation algorithm was proposed for solving the problem. In the solution process, multi-objectives were taken into account, including revenue, train load factor, passengers satisfied as well as computation time overhead. Finally, the model was validated by an example.(5) Finally, we proposed a seat inventory control model for multi-train under multi-price. Meanwhile, a column generation algorithm was designed to solve the model. Firstly, passengers'choice behavior was discussed under multi-price. Secondly, a dynamic seat inventory control model for multi-train under multi-price was established. Thirdly, a column generation algorithm was given to solve it. Fourthly, the model and algorithm was validated by an illustrative example. Finally, managerial insights were proposed for multi-price seats inventory control strategies.