Dynamic Vehicle OD Flow Estimation for Urban Road Network Using Multi-Source Heterogeneous Data

Dynamic OD flow plays an important role in transportation planning and management. In this paper, a dynamic vehicle OD flow estimation model of urban road network was developed using multi-source heterogeneous traffic flow data. First, in order to improve the accuracy of OD demand allocation, the GPS data, road topology, and land use attributes were considered to construct the network-level traffic zones. Second, the ALPR data and GPS data were combined to increase the accuracy of observable vehicle OD flow. Third, a Kalman filter model with linear state constraint was proposed to estimate the unobservable vehicle OD flow. Specifically, the state transition equation was developed using random walks. The observation equation with dynamic mapping relationship between OD flow and link flow was developed based on dynamic traffic flow distribution theory using data collected from microwave detectors and ALPR sensors. The linear state constraint was formulated with observed traffic demand of network-level traffic zones using ALPR data. Finally, the performance of the model was evaluated and analyzed with the field data of Kunshan, China. The results showed that the proposed model estimated link flows accurately and performed better than standard Kalman filter model.