The Optimal Toll Determination of Urban Expressway with Deep Learning Method

Since the distance based toll system has been introduced in urban expressways, the determination of optimum toll is quite important in terms of practical traffic management of urban transport networks. The traffic assignment with variable demand is usually applied to estimate traffic flows on urban network to evaluate the social benefit of pricing. However, the huge load of network calculation would be reduced to discuss the combination of parameters in the distance based toll. The deep learning approach is introduced to learn the essential factors in the estimation of traffic flows in urban transport networks. In particular, the convolutional neural network (CNN) is created from the database of the estimation results in the original example networks. The estimation model for the values of indicators is used in the determination of parameters in the distance based toll. The advantages of the deep leaning approach are to provide the approximate solution without traffic assignment process. It reflects on the efficient determination of shape of distance based toll corresponding to the variable OD conditions. Finally, the optimum solution for the specific condition of urban network can be determined practically. The proposed method would be applied to the determination of optimal distance based toll corresponding to the variable demand.