Origin-destination Estimation Based On Automatic Vehicle Identification Data

The significant increase of vehicles ownership leads to the more and more obvious contradiction of urban road traffic, and sometimes the occurrence of traffic congestion, which has great impact on the efficiency and safety of travel. Scientific and reasonable transportation planning, traffic management and the intelligent transportation system have become one of the important means to alleviate the urban traffic problem. Vehicle Origin-Destination Matrix(OD matrix) describes the traffic pattern, which can provide the necessary information for transportation planning. On the other hand, the OD matrix contains the vehicles’ driving information, working as the basis of making the different kinds of measurement of traffic management.Firstly, this study proposes a procedure to deal with the Automatic Vehicle Identification(AVI) data to get the turning movement, link count, real travel time and the partial vehicle OD information. Combining the calculation thought of Fratar of transportation planning, the population of OD matrix can be estimated by using the partial vehicle OD information. Secondly, the assignment matrix can be calculated based on only the turning movement, the turning movement and real travel time, the turning movement and the partial vehicle OD information. When using the partial vehicle OD information to get the assignment matrix, missing intersection data are repaired to get the more complete route. Thirdly, with the turning movement and link count as input data separately, we use the Generalized Least Squares Model of a formulation to minimize the sum of the deviation between the measured turning movement(or link count) and the estimated OD turning movement(or link count) and the deviation between the history OD matrix and the estimated OD matrix to estimate the OD matrix. Finally, under the two different circumstances: the real OD matrix known or the real OD matrix unknown, we evaluate the methods by using the Nguyen–Dupuis Network. The results show that the turning movement and partial OD information can improve the accuracy of OD estimation.After researching the methods mentioned above, the study estimates the vehicle OD matrix based on AVI data from Langfang. The results show that the procedure proposed can estimate the OD matrix efficiently. Finally, we can improve the accuracy of OD matrix estimation by considering the interval of researching time, the calculation of real travel time and t he raw data including other useful information.