Identification Method Of Traffic State On Urban Expressway Based On The Microwave Data

As the skeleton of the urban road network,the expressway plays an important role in satisfying long-distance trips of motor vehicle in city and improving the operational efficiency of the city.However,with the increasing of vehicle ownership,the phenomenon of urban traffic congestion becomes more and more serious.The expressway,which should have provided a higher driving speed for vehicles,can't play its good traffic function.Even during off-peak time,the phenomenon that vehicles for a while move for a while stop will appear.The smoothness of expressway determines the overall traffic condition of the city.Therefore,it is necessary to divide traffic situation of urban expressway reasonably and realize their own real-time traffic status judgment so as to master the traffic condition of expressway and improve the travel quality of residents.In this paper,based on a large number of expressway traffic flow data obtained by microwave detectors,the time-space characteristics of traffic flow parameters are analyzed.From the characteristics of expressway traffic flow,the fuzzy C-means algorithm based on initial state is used to analyze the traffic state,and the binary tree support vector machine is used to realize the real-time discrimination of traffic state based on clustering results.The main contents are as follows:First of all,combining with the latest research on traffic condition both at home and abroad,three traffic flow parameters,traffic flow,speed and occupancy,are selected to characterize the traffic state.In order to ensure the accuracy of the subsequent study,the anomaly data were identified by threshold method,traffic flow mechanism method and K-Neighbor Local Anomaly Test algorithm.Then,the anomaly data were repaired by adjacent data averaging method.the repaired data will be test by the value of LOF.Secondly,the temporal and spatial characteristics of the traffic flow parameters are analyzed and the macroscopic basic map are described,the initial state of the traffic state is defined by the elasticity definition which is applied on the curve of relationship between the occupancy and traffic flow,and the traffic parameter thresholds are obtained.Based on this,using fuzzy C-means algorithm to optimize the clustering results of each state,and the weight distribution of each parameter is taken into account in the clustering process.Finally,based on the clustering results,a binary tree is constructed by using the class divisibility,and on each leaf node of the binary tree the support vector machine is trained.Then the incomplete binary tree SVM model is obtained,and the real-time discrimination of traffic state is achieved.The discriminant results of this model are compared with those of neural network,the rationality of the model is verified.