Research On Passenger Demand Forecast Technology Of Railway Oriented To Operation Plan

In recent years,with the continuous expansion and development of the railway network in our country,the number of the long-distance trains has been enhanced.At the same time,the operation demand for overnight D-trains is also on the increase.Due to the high density of trains in the daytime,the comprehensive overhauling operation on high-speed railway always is processed from 0:00 to 6:00 at night,which is in contradiction with the running demand of overnight D-trains and results in the mismatch between the allocation of transport capacity resources and passenger transport demand.This thesis mainly studies the OD(i.e.,Origin-Destination)passenger demand of overnight D-trains,in order to improve the adaptability of train operation scheme to passenger demand and provide theoretical supports and optimization methods.Firstly,due to the distance and characteristic of passenger flow,a mixed logit model based on random nonlinear utility functions(RNUF-MXL)and a mixed logit model based on improved nonlinear utility functions(INUF-MXL)are established to maximize the overall utility value of passenger groups,respectively.According to Maximum Simulated Likelihood Method,the likelihood functions of two mixed logit models are taken as objective functions to reduce the computational complexity,respectively.At the same time,we adopt Metropolis-Hastings Algorithm to iteratively solve the probabilities of discrete random variables(i.e.,passenger travel cost and time)in each utility function of RNUF-MXL.After that,we estimate the unknown parameters and solve the optimal solutions of these two models by Improved Simulated Annealing Algorithm.The calculation time,convergence speed and iteration results are also compared with Newton Method,Ant Colony Algorithm and Simulated Annealing Algorithm.Then,the forecast results of passenger demand are taken as the input data of the integrated optimization model of stop plan and ticket allocation(SPTA-IOM)established in this thesis.The goal of this model is to maximize passenger satisfaction which is quantified as minimizing passenger travel time and railway revenue which is quantified as maximizing the passenger capacity.By analyzing the structural characteristic of the model,a heuristic algorithm based on Lagrangian Relaxation is used to solve the model to find out the optimal stop plan and the optimal allocation of tickets in each train section for improving the solving efficiency.Finally,because Beijing-Shanghai high-speed railway is the most representative one in China,we forecast the passenger demand and optimize the operation plan of overnight D-trains in this line.The rationality and applicability of the proposed models,included two forecast models and one optimization model in this thesis are verified.