Research On The Traffic Status Identification And Prediction Method For Urban Expressway

As the backbone road connecting all areas of the city,urban expressway bears a large proportion of transportation tasks in the city and plays an important role in the urban road system.Its traffic operation efficiency not only affects the normal commuting of the city,but also affects the development of the city.With the continuous progress of urbanization,the number of motor vehicles has increased significantly,and the frequent traffic congestion on urban expressways has become increasingly serious,and the congestion time has been increasing,which makes it impossible to play the efficient commuting function of expressways.In addition,there are abundant traffic collection devices on urban expressways,which can obtain massive dynamic traffic data and provide data support for traffic state identification and prediction.Therefore,it is of great significance for expressway traffic management and control to fully tap urban expressway traffic data resources and realize accurate identification and accurate prediction of expressway traffic state.In this paper,the urban expressway is taken as the research object,and the traffic state identification and prediction of urban expressway are deeply studied to provide decision-making basis for traffic management and control.The main work is as follows:(1)Traffic state identification method based on travel time extractionOn the basis of defining the traffic state classification standard of expressway,a traffic state identification method based on travel time extraction is designed.Firstly,the adaptive smoothing filter is used to reconstruct the space-time velocity field,and then the grid virtual vehicle method is used to extract the travel time of the road section,and then the traffic state identification process based on the travel time is designed.Finally,the validity of the method is verified by the measured data of an expressway in a megacity.(2)Traffic state identification method based on Transformer modelTraffic flow,speed,occupancy rate,the ratio of occupancy rate to traffic flow and the ratio of occupancy rate to speed are selected as the characteristic variables of traffic state,and a traffic state identification method based on Transformer model is designed.In the location coding,five feature variables are coded and location information is recorded.In the attention mechanism layer,the internal features of data are extracted by multi-head attention mechanism,and in the classification layer,the traffic state is recognized by softmax classifier.Finally,the measured data of urban expressway induction coil are selected for example analysis.(3)Traffic state identification method based on decision-level fusionAiming at the problem of insufficient accuracy of single traffic state identification method,a traffic state decision-level fusion method based on fuzzy comprehensive evaluation method is designed.The factor set and evaluation set are constructed,and the recognition rate of different traffic state recognition methods is taken as the reliability index,and the weight vector matrix of influencing factors is constructed.The membership degree is calculated by fuzzy operation,and the fused traffic state recognition result is obtained by using the maximum membership principle.Finally,the measured data of urban expressway are selected for an example analysis.(4)Multistep forecasting method of short-term traffic flow based on two-layer decompositionIn view of the complex characteristics of short-term traffic flow time series,the signal decomposition theory is introduced,and the traffic flow time series is decomposed by the empirical mode decomposition method of adaptive noise complete integration.The high-frequency components obtained by decomposition are decomposed twice by variational mode decomposition method,and the long-term memory neural network model is constructed for each modal component to make three-step iterative prediction,and the prediction results of each step are obtained by superposition.Finally,the measured data of urban expressway are selected to verify the multi-step prediction performance of the method.