Study On Pre-processing Methods Of Dynamic Traffic Data Based On SCATS

Intelligent transportation system (ITS) has been wildly used in urban trafficmanagement and control, and traffic information collection is the basis of ITS. Trafficdata such as volume from the detectors is very important to the implement andevaluation of all sorts of traffic programs. Currently, most data acquisition methods inurban traffic intelligent control system are using inductive loop detectors. Data fromloop detectors are easily missed or abnormal during the data collection, transmissionand processing process, which are affected by many elements such as equipmentfailure, burst fault in software and etc; therefore, a data preprocessing program isimperative to distinguish and recover these incorrect data. Limited by the detectionprinciple and influenced by the work environment of loop detectors, deviation iseasily generated in the detected data. Moreover, though the researches in data qualitycontrol models are various, the basic principle is remained as calculating anapproximate value of field data. No matter how advanced the incorrect dataidentification and revising are, negative effect will occur in traffic system operationwhen the incorrect data is used to the traffic control directly.Based on current data quality control models, data analysis and simulationanalysis methods are used to study the data preprocessing method of intelligent trafficsignal control system. First, based on the SCATS control principle and algorithm, theability to process data error is evaluated. Then, the traffic flow distribution regularityis researched by statistical analysis of the data, it is concluded that the traffic flowtime distribution rule and the spatial distribution rule based on multi-lanes flowdistribution by lanes, to provide the theory support for the data filtration and repairingmethods of the quality control model. By comparing the field investigated anddetector data, the data error distribution is been statistic analyzed, making up for thelack of detector data error research. Finally, the practicability and effectiveness of thedata filtration and repairing model is been verified through a case study.