Real-time traffic volume estimation with fuzzy linear regression

Accurate knowledge of future short time traffic volume on the travel network is critical for traffic signal control and guidance information. It is important and necessary to estimate traffic volume precisely in real-time with historic collected data. In this paper, based on literature review of several methods to estimate future short interval traffic volume, a fuzzy linear regression algorithm is presented to predict urban traffic flow dynamically, the model is put forward and the detail principle is studied. In contrast to other technologies, the fuzzy algorithm is able to integrate statistical volume and occupancy data and get the fuzzy correlation among them. The algorithm is tested with real traffic data and produces average estimates error of traffic volume with only 7%. Applied to Changchun Street in China for one year, the algorithm exhibits high performance in traffic volume estimation and play a big role in advanced traffic information system.