Research On The Short-Term Freeway Traffic State Forecasting Method Using The Multi-Source Data

In order to achieve traffic state accurate and effective forecasting, it must obtain a comprehensive, real-time, accurate traffic state dynamic information. There are two traditional traffic technology of information collection, fixed and mobile traffic information collection mode, each performance is excellent and is still the most commonly used current highway. In recent years, with the rapid development of mobile phone positioning technology, mobile positioning accuracy has been remarkable improvement in which the phone switches as traffic information collection technology has achieved more attention for its rich, massive amounts of data. At this stage, get rich, multi-source heterogeneous traffic data, and how to effectively use these data to accurately predict the future state of traffic information in a short time has become a hot topic.This paper describes the status of short-term traffic prediction important position and role in ITS, whereby the importance and necessity of state short-term traffic forecast studies, and research status of traffic state application of mobile phone data, techniques of multi-source heterogeneous data and traffic state forecasting methods are reviewed, detailing the traditional traffic and mobile positioning technology; next defines the concept of traffic state, and the extraction method of pretreatment and two space mean speed of traffic data for details; then, combined microwave detector data and mobile switching data, using SVM, to establish traffic state short-term forecasting model of single and multi-source data source data; followed, Xicheng Highway of Jiangsu to one section of as an example, simulating real scene, analog phones switched to obtain the corresponding data traffic, switching speed of the road link is predicted and select the evaluation of the model and verifing. The result show that the switched data of mobile phone in traffic state prediction is better than forecast results of microwave detector data, the forecast effect of multi-source data forecast model based phone switching data and microwave detector data is superior to any predictive model of a single data source. Finally, Jiangsu Provincial Expressway network operators and service intelligent platform as the background, made expressway traffic state prediction system based on multi-source data, and demand analysis, function, design and analysis of key technology framework for a detailed description, provide guidance for the construction of the intelligent platform.