Investigating effectively freeway service area inbound vehicles will help to manage the local traffic flow, and to improve the layout and service capacity of the freeway service area facilities. For the freeway service area without detecting devices, two new methods are proposed by using vehicles’ RFID (Radio Frequency Identification) data from freeway upstream and downstream roadside collecting units. The GMM-EM based method is proposed to obtain the probability distribution of the speed of passing vehicles. By the ant colony algorithm, the speed threshold for distinguishing vehicles entering or not entering the service area is gained. Then the variational inference Gaussian mixture clustering algorithm is established to realize the discrimination of whether the vehicle drives into the freeway service area. Using RFID data of vehicles from Chongqing Freeway, China, the experimental results show that the average accuracy of distinguishing freeway service area inbound vehicles is more than 96%.

Traffic engineering; Passing into service area; Kmeans-GMM-EM algorithm; Variational inference and Gaussian mixture clustering algorithm