【作者】 杨富社；

【导师】 陈宽民；

【作者基本信息】 长安大学 ， 交通运输规划与管理， 2015， 博士

【摘要】 近年来,随着我国综合国力的不断提升,城市公共交通事业得以飞速发展,但与国外先进国家和地区相比,尚存在较大差距,集中表现在信息化、智能化管理与应用水平低下,因此改善城市公共交通环境、实现城市公共交通的动态调度管理,具有重要的理论意义和实用价值。城市公交动态调度管理依托数字化信息采集、处理和网络通信技术,综合运用GIS(地理信息系统)、通信、网络、多媒体及计算机等技术,对城市公共交通系统的信息采集、存储与发布、运营调度及公交优先等进行网络化管理,实现客流变化和车辆运行状况的实时监测和公交资源的优化配置、合理调度。本文探讨了城市公交动态调度管理的结构框架,并就支持系统功能实现的相关技术方法进行了着重探讨,旨在通过对动态管理技术的研究、探索,为我国广大城市提高公共交通管理水平,改善居民出行环境,以解决城市交通拥堵问题,实现城市的可持续发展,提供一定的借鉴依据和参考。本文从城市公交客流的基本特征出发,以西安市的多条线路为分析对象,对大中城市公交客流的时间分布特征、空间分布特征和区域分布特征进行了分类总结,对城市公交线路的增长特征进行了阐述,为城市公交动态调度系统的构建提供了基本理论依据。应用公交IC卡数据对线路客流信息进行了分析、推算,并引入改进的灰色马尔可夫模型来预测线网客流信息,运用最小二乘法和有限差分得到了模型中参数a、b的待定向量,同时引入边值修正因子和马尔可夫精确化来减弱输入数据离散、波动过度的不利影响,提高了线网客流的预测精度。同时,从提高公交服务水平的角度,提出了面向乘客的公交动态信息的发布内容、发布载体和信息传递途径,分析了发布形式和信息更新方式。考虑道路条件、交通负荷、车辆因素、驾驶员等多种影响因素,从路段正常行驶时间、停车时间和延误时间三个角度入手,建立了基于BP神经网络的公交车辆行程时间预测模型。以西安市207路公交车为例,通过调用Matlab软件中的神经网络工具箱,并运用相关调查数据对网络进行训练,并运用检测样本对模型进行检验,将行程时间的预测结果和实际值进行对比分析,分析结果验证了模型的有效性和准确性,为面向乘客的公交动态信息发布系统和公交运营调度管理系统提供了支持技术和理论依据。在对公交调度系统模型和调度形式进行分析介绍的基础上,考虑公交线路站点乘客需求,以乘客候车时间延误最小及车厢内满载率较均匀为优化目标,建立了公交车发车时间表模型,并以车厢满载率为控制参数,对模型进行优化求解,确定高峰时刻的最优发车时间,同时提出了车辆动态调度的分析模型。以西安市13路公交车辆的采集数据对该模型进行有效性检验,分析结果表明应用该模型制定公交车辆发车时刻表,进行车辆运行调度,能够实现线路客流的均匀分布,有效降低乘客的候车时间成本,提高调度效率,并且在线路车辆不足的条件下,运用车辆调度模型进行车辆调度管理,可以快速恢复线路车辆的车头间距,有一定实用价值。最后,针对公交动态管理效益分析的实际需要,提出了直接效益和间接效益的分析指标,具体讨论了减少误乘损失、票价降低、出行时间节约、公交分担率提高、运输费用成本节约及减少交通拥堵所产生的综合效益,并以西安市为例,进行了案例分析,计算结果表明实施数字化信息管理效益显著,展示了良好的应用前景。更多还原

【Abstract】 In recent years, with the rising of China’s comprehensive national strength, urban public transport rapid development. Comparing with advanced foreign countries and regions, there still is a large gap. The information, intelligent management and application level is the concentrated expression. Improving the urban public traffic environment and realizing the dynamic dispatching management of urban public traffic have the important theoretical significance and practical value.Bus dynamic dispatching management system is an combination of digital information collection, processing and network communication technology, the integrated use of GIS(geographic information system), communication, network, multimedia and computer technology. It deals the city public transportation system of information collection, storage and release, operation scheduling and bus priority with network management to implement the passenger flow change and vehicle running status real-time monitoring and bus resources optimization configuration, reasonable scheduling. This paper probes into the structure of the urban bus dynamic dispatching management framework, and discussion of the support system function implement. Through the study of dynamic management technology, to improve the city public traffic management level in our country, improve the environment for residents, in order to solve the problem of urban traffic congestion. Realizing the sustainable development of city and providing certain reference and reference.In this paper, starting from the basic features of city bus passenger flow and Xi ’an several routes as analysis object, summering the time of large and medium-sized city bus passenger flow distribution characteristics and spatial distribution characteristics and classification of summarizing the characteristics of regional distribution. It provides the basic theory to construction of city bus dynamic scheduling system.Application of bus IC card data of route traffic information is analyzed and calculated. An improved grey Markov model is proposed to predict the network traffic information. Using the least squares method and finite difference model parameters in the backlog vectors, a and b at the same time introducing boundary value correction factor and Markov accurate to weaken the negative impact of the discrete input data, to improves the prediction precision of line network traffic.At the same time, from the angle of improve the bus service level in surface, proposing bus dynamic information content, the release carrier and the information transmission and analyzing the release form and way of information updating. Considering road conditions, traffic load, vehicle factors, drivers and various influence factors and the normal road travel time, stop time and delay time three aspects, establishing the model that based on the BP neural network of bus travel time prediction. Taking xi ’an bus 207 as example, use the neural network toolbox in Matlab software and apply the survey data to network training. And use test sample of travel time to test the model by comparing the actual travel time with predicted results. The analysis results verify the effectiveness of the model and provide theoretical support for dynamic information release system for passenger bus and bus operation scheduling manage.Based on the introduction and analysis of the bus scheduling system model and scheduling form, taking the minimum delay of passenger waiting time and relatively uniform of bus full rate as the optimization goal, established bus schedule mode. Taking carriage utilization as a controlling parameter, optimizing the model solution, determine the optimal departure time of rush hour, and put forward the analysis model of the vehicle dynamic scheduling. By testing the effectiveness of the model through the data collected from bus line 13, the analysis result demonstrate that the application of the model for bus timetable, bus vehicle scheduling, can achieve circuit uniform distribution of passenger flow, effectively reduce the passenger waiting time cost, improve the efficiency of the scheduling, and under the condition of insufficient line vehicles, the vehicle scheduling model for vehicle dispatching management, can quickly restore lines front of the vehicle spacing, has certain practical value.Finally, aimed at the practical need of efficiency analysis of dynamic public transport management, the direct and indirect benefits analysis indicators was put forward. And discussed the comprehensive benefits in detail including reducing the mistake by losses, ticket prices drop, travel time savings, the transit share rate lift, freight cost savings and reducing traffic congestion. Taking Xi’an city as an case analysis, the calculation results show that the implementation of digital information management benefit is remarkable, and shows a good application prospect.更多还原