Study On Identification Method Of Driving Behavior Based On Vehicle Infrastructure Integration

With the continuous development of automotive industry and a rapid increase in car ownership, traffic accidents and road congestion occur frequently, causing serious losses of personnel, property and environment. Traffic safety, traffic congestion and traffic pollution have become the three most difficult problems to be solved in transportation field. The development of wireless communication and sensing technology, and the arrival of big data era, both of them will make intelligent transportation system an effective means to solve the three major problems. As an important research direction of intelligent transportation, car-road collaboration can improve traffic safety, ease traffic congestion, and reduce environmental pollution by adopting advanced sensor technology, wireless communication and computing devices.Driving behavior is the main factor that causes traffic accidents and road congestion, identification of driving behavior is an important method for road traffic accident forecast in car-road collaboration. Based on the comprehensive analysis about research on driving behavior at home and abroad, first of all, the reason causing road traffic accidents is discussed on a macro level in this paper, combined with the theory of data mining. The data mining method of vehicle driving information is put forward, and the framework about forecasting system of traffic accidents based on car-road collaboration is built, which provides a theory and method basis for research on identification of driving behavior. Secondly, the microscopic traffic simulation model is established by using VISSIM software, and massive original simulation data is obtained and preprocessed by simulations of local road-network under different driving conditions. Then based on vehicle dynamic theories, the parameters of vehicle running status to the influence on traffic safety is analyzed, and the evaluation method for running status parameters is derived to complete the transformation from basic driving parameters to running status parameters. Finally, the identification model of driving behavior with parameters of vehicle running status as input is established based on BP neural network to realize the evaluation of driving behavior. By testing and improving the identification model of driving behavior, the results show that BP network model is more accurate on identification of driving behavior, which will provide certain reference for vehicle active safety and traffic accidents forecast.