Research On Traffic Data Processing Methods Based On Floating Car And WSN

With the development of technology and economy, and the increase in car ownership, the traffic situation has also become an important issue in people’s lives. Accurate traffic information is not only able to provide a reliable basis for the traffic control, but also an indispensable condition for traffic forecasting. Up to now, accessing traffic information can include many methods, such as WSN data, floating car data, et al. However, these methods have their own defects, such as the GPS accuracy having effect on collecting information of floating car, and therefore can lead to the collected data is not accurate enough.In this thesis, with analyzing the methods, advantages and disadvantages of floating car and WSN traffic information collection way, introduces traffic partition concept to design a regional collection model that combines the WSN and floating car, based on which the defects of floating cars and WSN traffic data collection are studied. With combining neural network, we designed a traffic information fusion model based on Elman neural network and a traffic state extracting model based on neural network. On the basis of regional center, this thesis fused the collected traffic information of floating and WSN and extracted traffic state information, and thus can get more accurate traffic data providing real-time and accurate traffic information for traffic control and traffic forecasting.Firstly, this thesis analyzed research status of traffic information collection and traffic data fusion technology, and then studied and designed a regional traffic information collection model according to the idea of regional traffic control, elaborated hardware structure of the model, software composition and topology. Next, this thesis analyzed the processing methods of traffic information collection of floating car and WSN, pointed out the defects and the necessity of traffic information fusion of floating car and WSN, and proposed a traffic information fusion model based Elman neural network according to the features of traffic data, and its validity is verified by experiments.Lastly, this thesis designed a traffic state extraction model based on neural network, and elaborated the theory and algorithm of this model, and performed simulation experiments on VISSIM platform with real urban vehicular traffic flow data of Dalian city, and constructed the traffic state extraction model with MATLAB. Experiments are conducted on the comparison and theoretical analysis.