Traffic demand prediction using ANN simulator

Abstract

The prediction of the traffic data is a vital requirement for advanced traffic management and traffic information systems, which aim to influence the traveler behaviors, reducing the traffic congestion, improving the mobility and enhancing the air quality. Both the stochastic time series (TS) techniques and artificial intelligent (AI) techniques can be used for this aim. Daily traffic demand in Second Tolled Bridge of Bosphorus, which has an important role in urban traffic networks of Istanbul has been predicted by both a TS approach using an autoregressive (AR) model, and an AI approach using an artificial neural network (ANN) model. The results have shown that the prediction error obtained by ANN model is smaller than the error obtained by AR model. The results have also pointed out that many other transportation data prediction studies can be implemented easily and successfully by using the developed ANN simulator. © Springer-Verlag Berlin Heidelberg 2007.