基于MD模型的公路节点客运量预测方法研究

发布时间：2018-04-13 04:27

本文选题：公路节点 + MD模型　；参考：《北京工业大学》2015年硕士论文

【摘要】：科学的运量预测对区域内各种客运方式的规划建设、运输组织、经济效益及市场分配等有巨大的影响,而现有的公路客运量预测方法多以短期预测为主,能够精细预测中长期的方法较少,且随着高铁客运的加入导致产生诱增运量及公路客运产生转移运量。为了能够合理预测,本论文对公路节点进行系统性研究,并对MD模型进行改善与深化,完善了其理论,建立了基于MD模型的公路节点客运量的预测流程。论文首先对国内外公路客运量的预测方法进行分析总结,对本文要采用的MD模型预测法的国内外研究成果进行阐述。其次,在“北京市城市交通运行保障工程技术研究中心”开放基金项目与“北京市公路客运枢纽站场规划布局基础研究”两个课题的支撑下,以北京市为例,对公路客运客流机理、需求结构及影响因素进行分析,提出公路节点的运营组织模式,并从定形、定性、定向、定量四个方面对公路节点进行系统性分析,为公路节点客运量预测奠定理论基础。再次,采用支持向量机、RBF神经网络、时序预测三种典型的预测方法对北京市公路客运量进行预测,对比各种方法的适用范围及优缺点,并对MD模型的适用性进行了分析。基于此,在MD模型的出行牺牲量模型中加入出行疲劳度和延误率两因素,通过追踪车辆的方法对这两个因素的相关参数进行了调查,进而构建新的出行牺牲量模型。针对出行者的行为时间价值,首次引入基尼系数来确定时间价值的方差,进一步改进及完善MD模型的理论与方法,建立了一套完善的预测流程。最后,以京津唐经济圈为例,进行公路节点客运量需求预测。与原MD模型和Nested-Logit模型进行对比,证明了改进MD模型的合理性及有效性。该研究对促进MD模型在我国公路客运量预测的推广及应用具有重要的意义。
[Abstract]:Scientific traffic forecasting has a great influence on the planning and construction, transportation organization, economic benefit and market distribution of various passenger transport modes in the region. However, the existing highway passenger volume forecasting methods are mainly short-term forecasting.There are few methods to accurately predict the medium and long term, and with the addition of high-speed rail passenger, the induced volume of passenger transport and the transfer volume of road passenger transport are generated.In order to forecast reasonably, this paper makes systematic research on highway node, improves and deepens MD model, perfects its theory, and establishes the forecasting flow of highway node passenger volume based on MD model.Firstly, this paper analyzes and summarizes the forecasting methods of highway passenger volume at home and abroad, and expounds the domestic and foreign research results of MD model forecasting method to be adopted in this paper.Secondly, under the support of the open fund project of "Beijing Municipal Transportation Operation and support Engineering Technology Research Center" and the "basic Research on Planning and layout of Beijing Highway passenger Transport Hub Station", taking Beijing as an example,The mechanism, demand structure and influencing factors of highway passenger passenger flow are analyzed, and the operation organization mode of highway node is put forward, and the systematic analysis of highway node is carried out from four aspects: fixed, qualitative, orientated and quantitative.It lays a theoretical foundation for highway node passenger volume prediction.Thirdly, support vector machine (SVM) RBF neural network and three typical forecasting methods of time series are used to forecast the passenger volume of Beijing highway. The applicability of MD model is analyzed by comparing the applicable range, advantages and disadvantages of these methods.Based on this, two factors, travel fatigue and delay rate, are added to the travel sacrifice model of MD model, and the related parameters of these two factors are investigated by means of tracking the vehicle, and a new travel sacrifice model is constructed.According to the behavioral time value of the traveler, the Gini coefficient is introduced to determine the variance of the time value for the first time, the theory and method of MD model are further improved and improved, and a set of perfect forecasting flow is established.Finally, take the Beijing-Tianjin-Tang economic circle as an example, carries on the highway node passenger volume demand forecast.Compared with the original MD model and Nested-Logit model, the rationality and validity of the improved MD model are proved.This study is of great significance to promote the popularization and application of MD model in highway passenger volume prediction in China.
【学位授予单位】：北京工业大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：U492.413

【参考文献】