铁路行包运量预测模型研究

铁路行包运量预测是以运输需求和内部供给为导向,综合考虑各种影响因素,对行包运量现状和发展的正确把握.探讨利用人工神经网络结合主成分分析的方法,建立铁路行包运量预测模型,解释并预测行包专列开行后铁路行包运量的增长趋势.实例分析的仿真结果表明,采用主成分分析法的广义回归神经网络模型结构简洁、预测精度高、收敛速度快,对相关铁路部门和企业的决策具有参考意义.     

“大敦煌”“大交通”背景下的现代艺术设计教学研究

兰州交通大学研究丝路文化、敦煌文化、研究交通的条件得天独厚,在这样的条件和背景下思考,研究建立一个的“大敦煌”、“大交通”现代艺术设计教学体系,有着重要的现实意义。     

多协议标记交换技术(MPLS)及应用

MPLS是当前最具发展前景的IP over技术,十分适用在骨干网与广域网中使用。IP和ATM技术的结合能充分发挥在流量管理和QoS方面的作用,其方式有叠加模式和集成模式,MPLS在帧中继和ATM运行的一个好处是它为这些面向连接的技术带来了IP的任意连通性。通过集成链路层(ATM、帧中继)与网络层路由技术,解决了Internet扩展和保证IP QoS传输的问题。     

Driver behaviour,truck motion and dangerous road locations – Unfolding from emergency braking data

Road vehicles equipped with measurement, computing, data storage and data communication capabilities can be utilised as probe-vehicles. Data received from such vehicles can provide valuable traffic and traffic safety information in respect of the covered routes and the connecting road network. In this study, trucks negotiating their normal daily haulage trips were used as probe-vehicles and the data recording their vehicular emergency events, such as abrupt braking events, detected by their on-board vehicular safety systems were analysed. The motivation for such an analysis is manifold. The aspect emphasized here is that the recorded vehicular emergency actions and events can be seen as surrogate safety events. Some of these surrogate safety events mark traffic incidents and therefore can be used as input by the surrogate safety assessment methodology. Though the vehicular emergency data used herein provides only sparse spatial and temporal coverage of the road network investigated, its analysis led to some interesting findings about interactions between drivers, trucks and roads.     

基于PCA-SVM的交通事件检测研究

随着自动控制等技术的不断发展,智能事件自动检测系统为高速公路事件检测提供了有效的途径。但传统的检测方法因为交通数据样本有限、输入交通特征过于冗余、参数设置无法达到最优等问题的影响而无法获得更优的性能。文中采用I-880实测数据集,运用主成分分析方法对交通输入特性进行特征选择,构建支持向量机事件检测模型,最后用粒子群算法选择支持向量机模型参数,获得了满意的检测效果,为设计高性能的事件检测算法提供了方法,具有一定的现实意义。     

从经济学角度缓解北京交通拥挤问题的研究

为改善北京城市交通拥挤现状,研究分析了北京机动车尾号限行、小轿车限购等政策,并采用了问卷调查、异地调研等方法。在异地调查和异地政策分析上,得以启发。使用SPSS软件对数据进行了专业、科学的统计分析。经编写组会商后,结果显示:经济学手段对缓解北京交通拥堵有较好的应用前景,编写组结合经济学对北京缓堵提出了建议。研究得到北京第二外国语学院韩景华教授的鼎力支持和辅导。     

Modelling driver propensity for traffic accidents: a comparison of multiple regression analysis and fuzzy approach

Abstract

This research proposes an assessment and decision support model to use when a driver should be examined about their propensity for traffic accidents, based on an estimation of the driver’s psychological traits. The proposed model was tested on a sample of 305 drivers. Each participant completed four psychological tests: the Barratt Impulsiveness Scale (BIS-11), the Aggressive Driving Behaviour Questionnaire (ADBQ), the Manchester Driver Attitude Questionnaire (DAQ) and the Questionnaire for Self-assessment of Driving Ability. In addition, participants completed an extensive demographic and driving survey. Various fuzzy inference systems were tested and each was defined using the well-known Wang-Mendel method for rule-base definition based on empirical data. For this purpose, a programming code was designed and utilized. Based on the obtained results, it was determined which combination of the considered psychological tests provides the best prediction of a driver’s propensity for traffic accidents. The best of the considered fuzzy inference systems might be used as a decision support tool in various situations, such as in recruitment procedures for professional drivers. The validity of the proposed fuzzy approach was confirmed as its implementation provided better results than from statistics, in this case multiple regression analysis.