Time-dependent OD is one of the key elements for dynamic traffic simulation and advanced traffic information system. However, the acquisition of time-dependent OD is a challenging due to the high requirement of the large-scale and time-varying traffic data. In this paper, a method of time-dependent OD estimation based on multi-source data fusion is proposed for expressway. License plate recognition data and traffic flow survey data are integrated in this research, and the time-dependent OD can be obtained synchronously. Firstly, the travel trajectory is classified according to the actual data, and the method to obtain OD is proposed by using the shortest path algorithm. Then, combined with the obtained OD and the main section flow survey data of the expressway ramp flow, the complete OD of the whole expressway is obtained through od inversion, and the time-varying distribution ratio of OD is obtained by road section flow. Finally, a dynamic traffic simulation model of Suzhou expressway is constructed by DTALite, and the traffic condition data from AutoNavi are utilized to validate the model performance. The mean absolute percentage error (MAPE) of speed is 7.9%, and the MAPE of traffic flow is 9.9%. The results show that the accuracy of the time-dependent OD estimation model is accepted, and the proposed method can help obtaining time-dependent travel demand for large-scale traffic network.

OD Estimation;Expressway;Multi-source Data;DTAlite