Multivariate calibration of single regime speed-flow-density relationships [road traffic management]

This paper presents a multivariate procedure for performing automated fitting of speed-flow relationships for different roads based on loop detector data. The procedure is shown to fit the observations for different freeway, tunnel and arterial data, thus demonstrating its flexibility in terms of representing different types of roads. Furthermore, the procedure also provides a fit that is reasonable for all data regimes, unlike many other single regime models that only fit free-flow or forced flow conditions data. Finally, this single-regime model provides a quality of fit that is consistent with most multi-regime models, without the need to deal with the complexities associated with the selection of regime break points. In addition to demonstrating the fit of the model to well known sample data from a standard traffic flow text books, fits to three different recent data sets with 1 to 5 minute loop detector data are also presented. These fits demonstrate that the flexibility of the proposed technique to deal with real-time data for both Europe and North America.