Seats Allotment Model Research Of Intercity High Speed Railway Based On Passenger Flow Forecasting

The short-term passenger flow forecasting and the reasonable model of seats allotment are the main parts for railway operation department management system's constructing.It has important theoretical and practical significance for the high-speed rail operations to be marketized,simplified and humanized,and for high-speed rail passenger transport to play a better role in the social and economic development.In this paper,given the inaccuracy of short-term,low occupancy rate and unfairness of seats allotment,these models,including the passenger demand forecasting model,pre-distribution and dynamic adjustment model for high-speed railways,are reconstructed based on the short-term passenger flow forecasting and seats allotment,then passengers' data of some high speed Railway is analyzed as instance to validate models.The main work of the dissertation is as follows:(1)By studying the characteristics of railway passenger flow data such as the time-varying,the nonlinear and the stochastic volatility,a combined passenger flow forecasting model combining wavelet packet decomposition with long-short term memory networks is proposed to solve the problem that traditional prediction model cannot predict the short-term passenger flow accurately.The passenger flow combination forecasting model is constructed by using wavelet packets to process information in the time domain and frequency domain of the increasingly refined information components in the analysis of mutation information,and by taking the advantages of long-short term memory models in generalization and high precision.Experiments indicate that the fitting degree of the combined model is 0.7832,which is20.38% higher than that of the seasonal model,and 44.66% higher than that of the long-short term memory model of empirical mode decomposition.Moreover,the normalized mean square error of the combined model is 0.4178,which is 31.16%smaller than seasonal model and 8.53% smaller than the long-short term memory model of empirical mode decomposition.In addition,the average absolute percentage error of the combined model is 0.1292,which is 6.24% lower than that of the seasonal model,which is 14.78% lower than that of the long-short term memory model of empirical mode decomposition.After analysis,it shows that composite model has better forecasting performance.(2)By researching the method of railway seats allotment,a passenger seats allotment model combining ratio predistribution and weighted polling dynamic allocation for high speed railway is put forward aiming at the matters that there are unfairness in some range,low revenue and low passenger-carrying rate with traditional method based on linear programming.First of all,using the combined forecasting method to obtain the passenger travel demand,the basic pre-distribution of the ticket amount is carried out according to the proportion of passenger flow to meet the minimum trip demand of the passenger flow.Then,the weighted round-robin allocation algorithm is used to dynamically allocate the amount of tickets for the remaining customers,so that limited number of tickets is more evenly allocated to the line section,and the train seats are utilized to the maximum extent to meet the passenger travel requirements in each section.Experiments show that the overall ticket occupancy rate of the dynamic ticket allocation model reaches 0.794,the average revenueof the single-train second-class seat reaches 114,130 RMB,and the fairness of seats allotment reaches 100%,compared with traditional linear planning ticket allocation model,these three property indicators mentioned above of the proposed model are increased by 8.67%,10.01%,and 10%respectively.