Predicting short-term travel time with considerable accuracy and reliability is critically important for advanced traffic management and route planning in Intelligent Transportation Systems (ITS). Short-term travel-time prediction uses real travel-time values within a sliding time window to predict travel time one or several time step(s) in future. However, the nonstationary properties and abrupt changes of travel-time series make challenges in obtaining accurate and reliable predictions. Recent achievements of deep learning approaches in classification and regression shed a light on innovations of time series prediction. This study establishes a series of Long Short-Term Memory with Deep Neural Networks (LSTM-DNN) layers using 16 settings of hyperparameters and investigates their performance on a 90-day travel-time dataset from Caltrans Performance Measurement System (PeMS). Then competitive LSTM-DNN models are tested along with linear regression, Ridge and Lasso regression, ARIMA and DNN models under ten sets of sliding windows and predicting horizons via the same dataset. The results demonstrate the advantage of LSTM-DNN models while showing different characteristics of these deep learning models with different settings of hyper parameters, providing insights for optimizing the structures.

Chapter Contents:

• 4.1 Introduction
• 4.2 Traffic time series estimation with deep learning
• 4.2.1 Recurrent neural network
• 4.2.2 Convolutional neural networks
• 4.2.3 Generative adversarial networks
• 4.3 The LSTM-DNN models
• 4.4 Experiments
• 4.4.1 Datasets
• 4.4.2 Evaluation metrics
• 4.4.3 Hyperparameter settings for LSTM-DNN models
• 4.4.4 Comparison between LSTM-DNN models and benchmarks
• 4.5 Conclusion and future work
• References

Inspec keywords: traffic information systems; regression analysis; autoregressive moving average processes; neural nets; traffic engineering computing; learning (artificial intelligence); time series

Other keywords: time series prediction; accurate predictions; deep learning models; travel-time dataset; sliding time window; deep neural network layers; competitive LSTM-DNN models; travel-time series; short-term travel-time prediction; travel-time values; long short-term memory

Subjects: Neural computing techniques; Other topics in statistics; Knowledge engineering techniques; Traffic engineering computing