Research On The Optimization Of Railway Ticket Pre-allocation And Dynamic Pricing

With the construction of China high-speed railway network,the volume of sending passenger has been increasing.The ticket sale decisions have become an important part of railway passenger transportation.At the same time,with the application of network technology,ticket reservation is becoming more centralized and real-time.However,there are some room for improving the management level of railway ticket sale at present.On the one hand,the passenger demand varies greatly between different lines and in different periods.The making and adjustment of ticket pre-allocation for different lines mainly rely on experience,lacking of theory and algorithm support which is accurately oriented to different scenarios.On the other hand,the price of railway ticket is based on operating mileage,which is not suitable for the diversified needs of travelers.Based on the theory and practice of revenue management in airline industry,this paper studies modeling and solution methods for railway passenger ticket sale in China,which has important theoretical and practical significance.The main contents of this paper include the following aspects:(1)Research on modeling framework of railway ticket sale problemIn order to consider the randomness of passenger arrival order and ticket purchase behavior,and to express complex railway seat inventory control mechanism,this paper proposes a fundamental model framework for railway ticket reservation problem,consisting of three parts: ticket sale decision model,ticket reservation process model and passenger behavior model.By applying Markovian decision process theory and discrete selection theory,the ticket reservation decision model and the passenger’s behavior model can be combined.In this framework,the seat inventory control mechanism is modeled by state transaction functions,which can be expanded flexibly.In the solution of the problem,due to the“curse of dimensionality”,this paper introduces the technique of approximate dynamic programming,and gives the general approach of solving the parameter adjustment model.The proposed modeling framework effectively provides the model expression of complex ticket sales rules and passenger behavior,which is the basis for analyzing and solving specific problems.(2)Research on optimization of ticket pre-allocationThe pre-allocation is a widely used type of inventory control in China’s railway.It includes a variety of seat control mechanisms such as sharing and reusing seat in a train.The key issue of pre-allocation optimization is to find a refined model to describe the preallocation mechanism.This paper expands the modeling framework to model the ticket pre-allocation mechanism,and then considers ticket allocation of the seat-sharing group.A genetic algorithm is used to solve the pre-allocation optimization problem,and a rulebased heuristic method is introduced to generate the initial solution of the problem.The feasibility of the solution results is verified through simulation experiments.The results show that the optimized ticket pre-allocation can significantly increase ticket sales when supply falls short of demand.(3)Research on dynamic pricing problem of high-speed railway based on dynamic bid priceThe dynamic pricing is an important mechanisms for the market operation of railway passenger transportation.And the dynamic bid price mechanism is one of the implementations of dynamic pricing methods.This paper studies the seat inventory control model based on dynamic bid price to determine dynamic price of railway tickets.Based on the modeling framework,the optimal control problem and value function approximation method are introduced.A dynamic programming model is proposed to calculate the bid price.In problem solution,this paper firstly transforms and simplifies the problem.In order to solve the large-scale linear programming problem,a dynamic aggregation method and a column generation method are proposed.The proposed solution method effectively reduces the number of variables and constraints.The solution test shows that the proposed method can accurately solve the bid price problem in a shorter solution time.In addition,the simulation experiment of ticket sales for passengers with different transfer intentions shows that the increase in passengers’ willingness to transfer will bring random fluctuations in ticket sales.(4)Research on railway dynamic price problem based on network flow modelIn the study ofrailway dynamic pricing problem,this paper also tries to adopt an new method based on network model,in which the passengers’ travel behavior and dynamic pricing mechanism are modeled as paths in the state network.By solving a network routing problem,a dynamic pricing strategy can be achieved in rid of the bid prices.To solve this kind of problem,this paper studies the Lagrangian heuristic method.By using the Lagrangian function decomposition method and the cutting plane method with trust region,the problem can be solved rapidly in parallel.The numerical experiment shows that,compared to the sub-gradient method,the proposed method can reduce the vibration of the Lagrangian multipliers in solution process and stably approximate the optimal solution.(5)Case study of Nanning-Guangzhou High-speed RailwayIn the background of the operation of Nanning-Guangzhou High-speed Railway in2016 summer,this paper tests different seat inventory control mechanisms including ticket pre-allocation and dynamic pricing in the varies demand scenarios.The results have shown that the flexible use of stock keeping unit(SKU)and flexible pricing mechanism can significantly increase ticket sales revenue.