基于GPS轨迹数据的超速事件检测方法实验研究

Recently, the detection of speeding events from traffic police department is mainly achieved by fixed-point speed measurement or interval speed measurement, but this method is not only large in error, but also difficult to detect all the vehicle speeding events. With the widespread use of GPS devices, some scholars attempted to use the speed information of GPS trajectory data to detect all the vehicle speeding events, but the single point speed is large in error. In order to detect all the vehicle speeding events in a promptly, accurate and comprehensive manner, the speeding event detection method based on N(N^ 2) consecutive speeding GPS trajectory points was studied through the driving test. The speeding event was considered as a linear event containing the starting and ending points and described by the speeding GPS trajectory points, and the driving test plan was designed based on that. Six roads in the suburbs of Chengdu were selected as test routes. The test vehicle traveled multiple times on each test route, using the vehicle-mounted GPS device to collect the driving trajectory of the test vehicle, and the on-vehicle personnel recorded the time and location of the vehicle speeding events. Based on the driving test data, the matching degree index was constructed, and the speeding event detection algorithm based on consecutive speeding GPS trajectory points was designed. The optimal consecutive speeding GPS trajectory points of speeding event detection were determined by comparative analysis, and the accuracy of the detection algorithm was validated. The results show that the criterion of speeding event is 5 consecutive speeding GPS trajectory points; the accuracy of the speeding event detection algorithm based on 5 consecutive speeding GPS trajectory points is 96%. The research results are able to lay a solid theoretical foundation for the dynamic monitoring of the speeding events, real-time presenting, the distribution of time and space revealing, contributing factors mining, and intervention methods proposing, which has a good practical application value.