Abnormal Data Recognition And Repair Of Traffic Flow On Expressway Based On Spatiotemporal Characteristics

In recent years,under the background of the formulation of medium and long term science and technology evolution,the implementation of urbanization strategy,the evolution of Belt and Road and the increasingly perfect urban road network structure,the urban traffic problem is becoming more and more prominent.Expressway plays a key role in the transportation industry and is an important carrier to realize the goal of long-distance fast travel in the urban road network structure.In the past ten years,because of the rapid development of national economy and the improvement of urban planning and construction in China,it is very important to solve the three hot issues of traffic development.The increase of vehicle ownership year by year not only greatly increases the burden on highway transportation,but also increases the incidence of accidents.The analysis of traffic flow operation state is affected by the accuracy of actual traffic flow data,and one of the technical bases of intelligent transportation system is detection data,which is directly related to the operation quality of the whole road traffic network and further affects the normal work of urban traffic system.Further exploration and full use of a large number of traffic flow data information can provide decision-making basis for urban traffic managers.This paper firstly analyzes the correlation of three parameters of traffic flow from the dimension of time and space,and establishes the three-dimensional ideal surface through the relation model of three parameters.Secondly,based on the traditional traffic flow theory,the chaotic model identification method is adopted,and the reconstructed surface is trained based on the grey GM(1,N)model,and the LSTM neural network and the RBF neural network are used to repair the abnormal data points.Finally,the traffic lane flow statistics measured by Interstate 5 Expressway in the United States are analyzed and verified by an example to verify the accuracy of the data identification and repair techniques in abnormal cases.In this paper,there are 32 charts,14 forms,and 61 references.