Research On Identification And Recovery Method For Abnormal Highway Traffic Flow Data

Highway is the main carrier to realize the long-distance transportation and occupies the pivotal position in Transportation industry. In recent years, due to the rapid economic development, the increasing number of motor vehicles not only increased the load of highway traffic but also the incidence of highway traffic accident. To improve the highway traffic security, the intelligent transportation system has been widely used. However, traffic flow data is the basis of this system, and the quality of traffic flow data can directly affect the analysis of traffic flow state. Therefore, it is necessary to improve the accuracy of traffic flow detector data.This paper aims at highway traffic flow data to identify and repair the abnormal data. Firstly, analyzes the time and space characteristics of traffic flow parameters, and according to the relational model among traffic flow, speed and occupancy data to establish the ideal surface of traffic flow. Secondly, for the identification of abnormal traffic flow data, proposes the identification method of highway traffic missing data based on mathematical statistics, and on the basis of threshold method and traffic flow mechanism, puts forward the identification method of erroneous data based on Chaos Theory; For the recovery of abnormal traffic flow data, this paper proposes a recovery method based on surface reconstruction, and this method adapts the RBF network to reconstruct the surface of scattered data for repairing the data, and will also compared with the recovery method of GM (1. N) model. Finally, using the measured Hong Kong and Macao highway traffic data to validate the validity and accuracy of the identification and recovery method put forward above.