Travel Time Prediction for Congested Freeways With a Dynamic Linear Model

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