Research On Urban Road Traffic State Prediction And Recognition

Traffic congestion is a serious problem facing major cities.Information support can be provided by fast and accurate traffic state prediction and recognition technology for traffic management,at the same time,it helps drivers plan suitable travel time to avoid congested routes.This paper mainly studies traffic state prediction and traffic state recognition.The main contents of this paper are as follows:(1)This paper expounds the traffic state classification index and the traffic state classification.In this paper,the average speed and traffic congestion index are selected as traffic state prediction indexes.Based on average speed and traffic congestion index of urban road in Beijing collected from Baidu map intelligent transportation platform,combined with cosine similarity,the similarity characteristics of average speed and congestion index in time dimension are analyzed,it provides a theoretical basis when the traffic state prediction model is established.(2)The traffic state prediction model based on adaptive network-based fuzzy inference system(ANFIS)is constructed,combined with the actual data collected,the model is verified to realize the direct prediction of traffic state.The average speed and traffic congestion index prediction model based on ANFIS are established.The indirect prediction of traffic state is realized according to the predicted value of two parameters.The direct prediction and indirect prediction based on ANFIS are compared and analyzed.The results show that the traffic state direct prediction model has good prediction effect.(3)The traffic state obtained based on the traffic parameters of the road section is the average value of the whole road section,but the traffic state of different places on the same road section is inconsistent.Therefore,aiming at the traffic state between different places,a traffic state recognition model is designed in this paper,it is built on the basis of convolutional neural networks.The transfer learning of the Inception V3 model is carried out,and some convolutional layers and fully connected layers of the Inception V3 model are modified.Based on this,the specific structure and parameters of the traffic status recognition model are determined.(4)The traffic status recognition model is built and trained by using high-level API keras,combined with the actual traffic image data,the model is verified.The evaluation index and confusion matrix are selected to analyze the verification results,it shows that the recognition accuracy of the model is higher,the model can accurately identify,model has strong recognition ability for urban road congestion.This paper has a good effect on the prediction and identification of urban road traffic state,and verifies the rationality and feasibility of the traffic state prediction model and recognition model,which can provide ideas and basis for alleviating urban road traffic congestion.The full text includes 51 figures,19 tables and 73 references.