[ 31 Dec 2021 | vol. 9 | no. 2 | pp. 75-82 ]

About Authors:

Daehyon Kim
-Department of Culture and Tourism Management, Chonnam National University, South Korea


Origin-Destination (O-D) flow estimation is one of the main topics of transportation planning and engineering research. Various research has been conducted for the estimation of origin-destination(O-D) flows from the link traffic volume over the past few decades. In addition, the artificial neural network model is a method that is attracting attention in various fields including transportation engineering and planning. More importantly, it is known that artificial neural network models are much more reliable and efficient than conventional models. In this study, O-D estimation was performed from the real-world link traffic volume data in the current highway network using an artificial neural network model. In particular, studies on network architectures and activation functions for learning were included to obtain improved results in the predictive accuracy of neural network models. In this study, a learning method to apply a neural network model to O-D estimation was proposed. Experimental results showed that the proposed neural network model can be much more efficient and more accurate than the existing general methods. In particular, the experimental results of this study showed that the artificial neural network model for real-time traffic estimation can be applied to dynamic traffic assignment.


Origin-Destination (O-D) Flow, Artificial Neural Network, O-D Estimation, Link Traffic Volume, Real-time Traffic Estimation


About this Article: