Hybrid empirical mode decomposition-neuro model for short-term travel time prediction on freeways

Accurate short-term prediction of travel time is central to many ITS systems, especially for ATIS and ATMS. In this study, we propose an innovative methodology for such prediction. Although the model can be theoretically used to predict traffic conditions using any of the three primary detector-based traffic parameters, the study was limited to the use of speed only as a single predictor. This was justified by the inherently direct derivation of travel time from speed data.; The proposed method is a hybrid one that combines the use of the Empirical Mode Decomposition (EMD) and a multilayer feedforward neural network with backpropagation. The EMD is the key part of the Hilbert-Huang Transform, which is a newly developed method at NASA for the analysis of non-stationary, nonlinear time series. The EMD is a straightforward to implement and computationally efficient method that is used to decompose any time series into a small number of its basic components, called the Intrinsic Mode Functions (IMFs). The rationale for using the EMD is that because of the highly nonlinear and non-stationary nature of link speed series, by decomposing the time series into its basic components, more accurate forecasts would be obtained.; We demonstrated the effectiveness of the proposed method by applying it to real-life loop detector data obtained from I-66 in Fairfax, Virginia. The method was used to predict link speeds for one through five periods ahead using 5-minute intervals across the eastbound direction of this corridor. To ensure proper testing, the data was compiled from different days with a wide range of traffic conditions, ranging from free-flow states to heavy congestion states. The prediction performance of the proposed method was found to be superior to previous forecasting techniques: conventional ANN, real profile, and historical profile. Rigorous testing of the distribution of prediction errors revealed that the model produced unbiased predictions of speeds. The superiority of the proposed model was also verified during peak periods, midday, and night.; In general, the method was accurate, computationally efficient, easy to implement in a field environment, and applicable to forecasting other traffic parameters. However, the proposed model requires additional effort on the part of the modeler. It also should be noted that the technique requires larger memory size for input feature expansion resulting from the EMD compared with a conventional ANN. Moreover, more testing of the effectiveness of the method under non-recurring congestion is recommended.