Research And Development Of Real-time Traffic Estimation System Based On Data Driven

To solve urban traffic congestion caused by the continuous growth of private passenger car ownership in China,on the one hand traffic infrastructure needs to be built to improve road capacity,one the other hand large cities need more traffic guidance to guide the public to travel reasonably.Real-time vehicle information is collected to identify the traffic status of the road section and then the traffic information can be published to users through various channels,which can achieve a reasonable induction of public travel routes.Most of the current research focuses on the use of GPS floating cars and roadside unit to obtain real-time traffic information,which is lack of full use of data generated by a single running vehicle on the road.With the popularization of intelligent connected vehicle,using intelligent connected vehicle as a floating vehicle to obtain real-time data of running vehicles can identify traffic status more accurately.In this thesis,a data-driven real-time traffic estimation system is put forward and four components of the system on-board terminal,traffic estimation cloud platform,traffic display app and management system are designed and developed.The on-board terminal communicates with vehicle network via OBD-II interface and keep communication with the cloud platform by its wireless communication module.The traffic estimation cloud platform is responsible for processing and calculating vehicle data to get traffic information,providing data query interface to the traffic display app and management system.The function of the traffic display app is providing traffic query service to users and allowing car owner users to remotely control their vehicles on their phones.In order to help traffic estimation system administrators manage the system more efficiently,the management system is developed to manage connected vehicles,on-board devices and users.In this thesis,the support vector machine algorithm is used to calculate and estimate the real-time traffic in the cloud platform,and a large number of vehicle simulation data generated by the micro traffic simulation platform SUMO is used to validate the algorithm’s estimation accuracy in different connected car penetration.The result shows that the estimation accuracy can reach 85% when the connected car penetration is over 20%.The data-driven traffic estimation system developed in this thesis combined with support vector machine algorithm can accurately identify realtime road conditions in higher connected car penetration,and users can get the traffic estimation results via the system’s traffic display app,which have certain practical value.