Research On Urban Road Traffic Incident Detection Method Based On Big Data Of Vehicle Trajectory

In the era of rapid development of traffic big data applications,Intelligent Transport System has made rapid progress and development,more and more researchers use data more effectively and intelligently in the field of urban traffic management and control.The use of trajectory data for event detection provides effective data support and theoretical support for traffic managers to solve a series of problems.Such as traffic safety or environmental problems brought about by traffic events.In previous studies,there were many problems such as inadequate analysis of traffic conditions,complicated detection procedures,and higher detection errors.In order to solve these problems,this paper analyzes and studies the vehicle trajectory big data,and proposes a method based on floating vehicle for rapid detection of urban traffic events and analysis of event types.The goal of the method is to use the trajectory data to detect sporadic or frequent events occurring during the running of the vehicle,thereby ensuring the timeliness and accuracy of traffic incident detection.The experimental data of this paper is taken from Beijing taxi data,first of all,performed data preprocessing,combined with road network information,map-matching is completed by the data point,position information and direction angle.Secondly,according to the characteristics of typical traffic flow in each road section,taking into account the changes of intersections and bearings,at the same time judging the direction of vehicle and recreating the traffic trajectory from the trajectory data.The trajectories are classified to solve some precision error problems caused by tunnel crossing and abnormal driving,so as to improve the accuracy of trajectory reproduction.Based on this,a GIS-based track query system is made.Finally,by analyzing the characteristics of traffic parameters when traffic events occur,an algorithm model for detecting event occurrence is established.Then,an improved Cumulative Sum Control Chart(CUSUM)algorithm is proposed.The method determines the event occurrence point by analyzing the characteristics of the traffic parameters and checking the data offset during the event occurrence.The track recurring system reproduces the trajectory of the event,provides a reference for judging the cause of the event,and combines the map data to characterize the type of the event.The detection accuracy is verified by VISSIM simulation.The detection results are compared with the traditional California algorithm and BP neural network algorithm by the detection rate,false positive rate and average detection time.The detection accuracy and efficiency of the method are more efficient.In addition,this method can also judge the type of the event well.