基于组合预测模型的交通流预测研究

发布时间：2018-11-20 16:56

【摘要】：城市交通系统担负着运输城市中人流物流的责任,是一个城市的血脉,其运行效率直接影响着整个城市。近年来,随着社会经济的发展和城镇化的不断深入,城市人口和机动车保有量迅速增长,虽然城市交通系统在基础设施方面不断建设和完善,但还是远远赶不上城市对于交通资源的需求。随之产生的交通拥堵问题、交通污染问题给城市企业和居民的日常生产生活带来了极大的不便,造成大量社会资源浪费。ITS(Intelligent Transport System)智能交通系统是21世纪的一种综合交通管理系统。其依靠对道路中行驶车辆运行状态的监控、控制、诱导,优化交通流在城市路网中的合理分布,提高城市路网的使用效率,从面上提高整个城市的交通系统运输效率。智能交通系统的两个重要组成部分即交通诱导和交通控制。而交通流诱导和控制的前提,是对城市路网交通做到合理的预测。本文的主要工作是,设计一种交通流预测组合模型,提高交通流预测精度。文章在对城市交通流特性分析的基础上,提出一种满足城市交通流线性、非线性、突发性的交通流组合预测模型。模型的基本思路是,先对交通流数据采用单模型进行预测,其中以ARIMA模型体现交通数据的线性,神经网络部分体现交通数据的非线性,K邻近非参数回归模型体现突发性,再运用神经网络拟合预测单模型的预测结果。在上述研究的基础上,运用组合模型重点研究了不同时间间隔的交通流预测和不同时段的交通流预测两部分内容。使模型在不同预测需求条件下都能准确预测,增强模型的适用性。运用IBM SPSS Modeler软件进行建模,对明尼苏达大学德卢斯分校的交通数据研究实验室提供的实验数据进行分析预测,最终得到了组合模型能取得优于一般单项模型精度的结果。
[Abstract]:Urban transportation system bears the responsibility of people flow and logistics in transportation city, and it is the blood of a city, and its running efficiency directly affects the whole city. In recent years, with the development of social economy and the deepening of urbanization, the number of urban population and motor vehicle has increased rapidly, although the urban transportation system has been continuously built and improved in terms of infrastructure. But it still falls far short of the city's demand for transportation resources. The traffic congestion problem and the traffic pollution problem have brought great inconvenience to the daily production and life of urban enterprises and residents. . ITS (Intelligent Transport System) Intelligent Transportation system is a comprehensive traffic management system in the 21st century. It depends on monitoring, controlling, inducing and optimizing the distribution of traffic flow in the urban road network, improving the efficiency of the urban network and improving the transportation efficiency of the whole urban transportation system. Traffic guidance and traffic control are two important components of Intelligent Transportation system (its). The premise of traffic flow guidance and control is to make a reasonable prediction of urban road network traffic. The main work of this paper is to design a combined traffic flow forecasting model to improve the accuracy of traffic flow prediction. Based on the analysis of the characteristics of urban traffic flow, this paper presents a combined forecasting model of urban traffic flow, which satisfies the urban traffic streamline, nonlinearity and paroxysmal. The basic idea of the model is that the traffic flow data is first predicted by a single model, in which the ARIMA model is used to represent the linear traffic data, the neural network partly reflects the nonlinear traffic data, and the K-neighborhood nonparametric regression model reflects the paroxysmal. The prediction results of single model are fitted with neural network. On the basis of the above research, the combined model is used to study the traffic flow forecasting at different time intervals and the traffic flow prediction at different time intervals. The model can be accurately predicted under different forecasting demand conditions, and the applicability of the model can be enhanced. The IBM SPSS Modeler software is used to model the experimental data provided by the Traffic data Research Laboratory of the University of Minnesota at Duluth. The results show that the combined model can achieve better accuracy than the general single model.
【学位授予单位】：兰州交通大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：U491.112

【相似文献】