Travel time prediction for congested freeways with a dynamic linear model
This work addresses the problem of accurate travel time prediction for Intelligent Transportation Systems, offering an incremental improvement over existing methods like k-nearest neighbor and neural networks.
The paper tackled travel time prediction on congested freeways by proposing a dynamic linear model (DLM) that adapts parameters over time to approximate non-linear traffic states, resulting in significant accuracy improvements, especially for short-term predictions, as demonstrated on California freeways I210-E and I5-S.
Accurate prediction of travel time is an essential feature to support Intelligent Transportation Systems (ITS). The non-linearity of traffic states, however, makes this prediction a challenging task. Here we propose to use dynamic linear models (DLMs) to approximate the non-linear traffic states. Unlike a static linear regression model, the DLMs assume that their parameters are changing across time. We design a DLM with model parameters defined at each time unit to describe the spatio-temporal characteristics of time-series traffic data. Based on our DLM and its model parameters analytically trained using historical data, we suggest an optimal linear predictor in the minimum mean square error (MMSE) sense. We compare our prediction accuracy of travel time for freeways in California (I210-E and I5-S) under highly congested traffic conditions with those of other methods: the instantaneous travel time, k-nearest neighbor, support vector regression, and artificial neural network. We show significant improvements in the accuracy, especially for short-term prediction.