Research On Urban Road Intersection Safety Early Warning

Urban road intersection is an important node of the urban road traffic system, and it is also an important part of the urban traffic at the micro-level. The intersections can connect different directions’ road to form a road network, so vehicles can run in the road network. In the urban road intersection, a variety of traffic flow includes motor vehicles, pedestrians, non-motor vehicles imports from all directions, making the intersection traffic volumes increase substantially, intersection traffic chaos and reducing the efficiency of the urban road, is more likely to result in traffic accidents. Therefore, we must strengthen the study of the urban road traffic safety early warning, find the law of the accident, and then make predictions on the urban road traffic safety. Once an intersection abnormal accident, the traffic safety management departments can base on the safety alarm to optimize the geometric design, road conditions of the intersection, then to reduce intersection accidents and improve the intersection safety level.A neural network is used in this thesis to forecast the urban road intersection safety incidents. Based on the early warning theory to obtain the urban road intersection safety early warning flow, while grasp the relationship between safety early warning and forecasting of urban road intersections. Through the study of domestic and foreign road traffic safety microscopic forecast method, the thesis selects the neural network method to establish urban road intersections safety prediction model. Through comprehensively analyze the urban road intersection, establish safety early warning index system of the urban road intersection and select eight main factors as input parameters of neural network model, the number of accidents is the output parameters of the neural network model. Finally establish urban road intersections safety prediction model on the basis of RBF neural network and BP neural network. The instance analysis is carried out with statistical data of Beijing road intersections, and use MATLAB simulation to solve the model. The predicted results show that the prediction accuracy of the urban road intersection safety prediction with using RBF neural network is higher than using BP neural network. Finally, the accident prediction results of the RBF neural network is analysed, and while in different emergency, the corresponding safety measures should be taken by traffic management department.