Dynamic OD Prediction for Urban Networks Based on Automatic Number Plate Recognition Data: Paramertic vs. Non-parametric Approaches

Abstract

OD flows provide important information for traffic management and planning. In this paper, we propose four OD prediction models based on the data obtained by Automated Number Plate Recognition (ANPR) cameras. The principal component analysis (PCA) is applied to reduce the dimension of the original OD matrices and to separate the main structure patterns from the noisier components. A state-space model is established for the main structure patterns and the structure deviations, and is incorporated in the Kalman filter framework to make prediction. We further develop three K- Nearest Neighbor (K-NN) based pattern recognition approaches. The proposed four approaches are validated with three days’ field ANPR data from Changsha city, P.R. China. The results show that on one hand our proposed approaches are able to make accurate prediction of OD flows under different demand conditions. On the other hand, the prediction accuracy is highly dependent on the quality of the available OD data: the Kalman filter model performs better for regular and periodic OD patterns; while for irregular OD matrices K-NN models could make more accurate prediction.