Research on vehicle behavior prediction based on high precision trajectory data

In order to monitor the driving behaviors of vehicles on urban roads and give early warning of dangers, the control system needs to identify the behaviors of vehicles in a short time and guide the future driving trend. Based on the machine learning method, the vehicle behavior characteristics are extracted. The vehicle behavior prediction model based on high-precision trajectory data is established to recognize and predict the vehicle lane changing and car following behaviors. The research uses the binary logistic regression method to analyze the traffic parameters between vehicles, analyzes the traffic information such as vehicle speed, vehicle head angle, relative position, and the relative speed with surrounding vehicles, and obtains the influencing factors of vehicle behaviors. This study establishes a vehicle behaviors prediction model based on BP neural network model. The results show that 12 factors are strongly correlated with vehicle behavior. The behavior prediction model can accurately predict the left-right lane change and car following behavior of vehicles, and the comprehensive prediction accuracy of the model can reach 93.9%. The research provides a theoretical and data basis for intelligent transportation development and urban road traffic management.