Pre-allocation Strategy Adjustment Based On The Expected Return

With the development of national economy, railway traffic continues to grow, and the railway transport capacity get released. At present, our country’s railway transportation is transferring from extensive to refinement in order to mine the potential for further transportation and increase the transportation efficiency. As a key work in the railway passenger transport organization and management, ticket distribution optimization method became the main way to realize the potentialities increase.In recent years, the ticket preassigned plan as supplement for ticket distribution basic plan was introduced to realize ticket allocation optimization when passenger flow fluctuate largely. The ticket preassigned plan depends on the precision of prediction, when accurate prediction can achieve good effect. These forecast methods are influenced by outside factors, and there are always some errors in these prediction results. When we use these methods to predict all ODs’ passenger flow value, calculation error will be magnified many times on train’s earning. This paper will try to use probability algorithm to solve OD’s passenger flow value after knowing the passenger traffic volume of this train, to get the optimal ticket preassigned plan.The main research content includes follows.1,Organize ticket allocation method and ticket preassigned business, describe the definitions, principles, methods and process of preassigned plan in detail. On this basis, analyze the ticket preassigned business requirement and constraints.2,Do research on the random distribution of passenger flow based on historical data. Focus on these three categories:daily passenger flow, holidays passenger flow, passenger flow of the same OD during the same time period. We use three methods like diagram of passenger flow, passenger flow value frequency chart, and data index to analyze these three passenger flow. And we prove that passenger flow is similar over the same period time in different years from aspects like the general trend, the value distribution interval and the fluctuation range of passenger flow.3,Base on the similar passenger flow rule over the same period time in different years, we use Monte-Carlo methods to analyze passenger flow data over these years, get the probability of different passenger flow value, and finally establish the ticket preassigned model based on expected return. When solve the preassigned model, the paper mainly studies how to solve the actual traffic distribution probability under huge amounts of data, and how to effectively reduce the possible preassigned strategy collections. At last, we do experiment on specific train, and demonstrate the availability of the preassigned model and present the process through this experiment.