Recognition Of Abnormal Traffic Behavior On Road Section Based On Sparse Floating Car Data

With the development of intelligent transportation system(ITS),the way people get traffic data is more efficient and convenient.The application of floating car technology in urban intelligent transportation system is becoming more and more mature.By analyzing the floating car data,the road traffic status can be obtained in real time.However,for some urban non critical intersections with imperfect traffic equipment or sparse floating car data sections with small traffic flow,the abnormal traffic behavior of individual drivers will cause traffic state error presentation.Based on the floating car data in Beijing,this paper improves the road map matching algorithm on the basis of analyzing and processing the floating car data,and uses the road grid generation method to generate the speed matrix to complete the missing data of sparse road sections.On this basis,the recognition model of abnormal traffic behavior is established and the algorithm is verified.First of all,the collected GPS data are preprocessed,and the solutions to the abnormal data such as data redundancy and data missing are proposed.After cleaning up the original data,the road grid method is adopted to improve the efficiency of road matching.Then,the vehicle GPS trajectory data is combined with the road geometric features to divide the road grid blocks to generate the speed matrix.The sparsity problem of the data is effectively solved by using schatten-p norm matrix.The low rank matrix data recovery method based on schatten-p norm can reduce the speed estimation error when the data availability is low.In this paper,the matrix completion algorithm is applied to the road traffic speed prediction,and the completion effect of different completion algorithms such as mean value method and singular value decomposition method is compared through the experimental evaluation index.Finally,based on the completed speed matrix,the road grid traffic state is demarcated,and the abnormal traffic behavior recognition model under the condition of sparse data is established.The model is solved by BP neural network,and the algorithm is verified by the actual data.The improved schatten-p norm matrix completion algorithm in this paper has a more accurate completion effect on missing data.The abnormal traffic behavior recognition model can effectively identify the abnormal traffic behavior of road vehicles,effectively reduce the error of road traffic state discrimination,and provide more accurate road traffic state information for traffic travel.