Affiliations: School of Transportation, Wuhan University of Technology, Wuhan, China
Correspondence:
[*]
Corresponding author. Wenbin Xiao, School of Transportation, Wuhan University of Technology, Wuhan 430063, China. E-mail: [email protected].
Abstract: In order to overcome the inaccuracy of current research results of traffic flow prediction, this paper proposes a prediction method for traffic flow with small time granularity at intersection based on probability network. This method takes one minute as time granularity, collects traffic data such as cross-section flow, section traffic flow velocity data, traffic density, road occupancy, section delay and steering ratio by using RFID technology, and analyzes and processes the data. By introducing Bayesian network in probabilistic network and combining K-nearest neighbor method, historical data and predicted traffic flow state are classified to realize the prediction of traffic flow with small time granularity at intersections. The experimental results show that this method has high prediction accuracy and reliability, and is a feasible traffic flow prediction method.
Keywords: Probabilistic network, intersection, small time granularity, traffic flow
prediction, bayesian network
