Research on Risk Assessment Technology of National Highway based on Deep Neural Network
In order to solve the difficulty of adapting the iRAP road risk assessment model to the risk assessment under different road environment conditions in the practical application of engineering,the basic information data of some national highway roads in the three eastern coastal provinces of China are collected,and a risk assessment dataset is obtained by video image processing,manual information coding and expert calibration.On the basis of the systematic framework of the original model of iRAP,the deep neural network is used to adjust and update the weight coefficients of the model,so as to construct the road traffic safety risk assessment model NNRAP on the typical road sections of national highways.The results show that the NNRAP model exhibits good performance in multi-classification road risk classification,with a classification accuracy of 90.35％on the test set;and in dichotomous road risk classification,the NNRAP model achieves a classification accuracy of 94.08％for road high-risk road sections,which is significantly better than the classification accuracy of the iRAP model.
traffic engineeringroad risk assessment techniquesdeep neural networkweighting factorNNRAP model
NETL
NSTL
万方数据
