Urban Road Traffic Congestion Detection Using RFID Data of Vehicles

Abstract

In recently years, an increasing number of vehicles cause heavier traffic congestion problems. The detection of traffic congestion has become a hot research topic around the world. Existing traffic detection methods mainly employ video data and GPS data, which are limited by their own characteristics. Video data are easily effected by weather conditions. GPS data only provide information of some vehicle and reflects partial traffic information. With the development of RFID technology, traffic data collection for all vehicles has been realized. This paper proposes a traffic congestion status detection model for urban road segments based on RFID data of vehicles. First of all, reasonable traffic flow parameters are selected to describe traffic congestion status, and a qualitative definition of traffic congestion levels is presented. Then, we develop a method to match traffic congestion levels with cluster centroids extracted by Fuzzy C-means(FCM) clustering. To automatically identify clusters, an iterative Fuzzy C-means clustering method is utilized. We finally introduced Generalized Equilibrium Fuzzy C-means(GEFCM) clustering to deal with the unbalanced sample data. The performance of the proposed model is investigated with vehicle RFID data from two key bridges in Chongqing. The model results show GEFCM outperforms FCM in term of accuracy, however, GEFCM owns high time consumption. We choose the clustering result of GEFCM when its fuzzy coefficient is 2.0 as the best one to perform traffic status analysis.