Spectral basis neural networks for real-time travel time forecasting

This paper examines how real-time information gathered as part of intellige nt transportation systems can be used to predict link travel times for one through five time periods ahead (of 5-min duration). The study employed a s pectral basis artificial neural network (SNN) that utilizes a sinusoidal tr ansformation technique to increase the linear separability of the input fea tures. Link travel times from Houston that had been collected as part of th e automatic vehicle identification system of the TranStar system were used as a test bed. It was found that the SNN outperformed a conventional artifi cial neural network and gave similar results to that of modular neural netw orks. However, the SNN requires significantly less effort on the part of th e modeler than modular neural networks. The results of the best SNN were co mpared with conventional link travel time prediction techniques including a Kalman filtering model, exponential smoothing model, historical profile, a nd real-time profile. It was found that the SNN gave the best overall resul ts.