Study on Traffic Detector Data Cleaning and Characteristic Analysis

With the development of collecting methods and information technology, more traffic detector data can be used to study traffic phenomena. However, the direct use of original data with quality problems to the transportation applications would result in unstable and unsafe application results. Therefore, it is necessary to develop an effective model for road traffic flow data cleaning and repair, so as to obtain the correct traffic information and grasp the law of traffic demand. This study, concentrating in Shanghai urban road intersections, uses SCATS and FCD data as sources and establishes a cleaning and repair model to preprocess data from detectors. The precision of the model is validated by corresponding data. By displaying the characteristics of the running speed and flow in urban roads, we find that the error of the model can be controlled at a reasonable level.