A Nonlinear State Space Approach to Arterial Travel Time Prediction

Abstract

The study uses time series and the Kalman prediction techniques along with modern technology such as the Global Positioning System (GPS) for accurate data collection and analysis. A greater understanding of travel time will help facilitate traffic system performance monitoring, control, planning, and informed route decisions for motorists accessing information from changeable message sings (CMS). The models used for estimations include the autoregressive integrated moving average (ARIMA) and the autoregressive moving average (ARMA). The study collects travel data for the peak hours of travel (3:30-5:00 p.m.) over an eight-month period on the busiest section of Highway 194 in Duluth, Minnesota. The predictions were conducted over two weeks during the summer of 2005. Observed and predicted travel times are charted carefully and report evaluations determine the success of the study.