Traffic density estimation with the cell transmission model

Abstract

A macroscopic traffic flow model, called the switchingmode model (SMM), has been derived from the cell transmission model (CTM) and then applied to the estimation of traffic densities at unmonitored locations along a highway. The SMM is a hybrid system that switches among different sets of linear difference equations, or modes, depending on the mainline boundary data and the congestion status of the cells in a highway section. Using standard linear systems techniques, the observability and controllability properties of the SMM modes have been determined. Both the SMM and a density-based version of the CTM have been simulated over a section of 1-210 West in Southern California, using several days of loop detector data collected during the morning rush-hour period. The simulation results show that the SMM and CTM produce density estimates that are both similar to one another and in good agreement with measured densities on 1.210. The mean percentage error averaged over all the test days was approximately 13% for both models.

Piecewise-Linearized Cell Transmission Model and Parameter Calibration Methodology

“…It can be seen that the densities and flows predicted by the models are similar to one another and agree reasonably well with the measured values, although the models tended to overestimate density during the peak congestion hours. These results are very similar to those of the CTM and SMM validation tests of (6,7), in which hand-tuned parameters were applied to the same freeway section. A more detailed description of the section geometry, cell partition, and data processing methods can be found in (6, 7).…”

“…The values of MMPE are not surprising, since they are similar to the MPEs in the short-segment validation tests of (6). From Figure 5 and Table 2, it can be seen that the CTM reproduces the observed bottlenecks and the approximate duration and spatial extent of the congestion upstream of each bottleneck, and predicts the total travel time with approximately 2% mean error over three days.…”

“…The CTM and SMM were validated on a 2 mile (3km) segment of I-210W, and shown to perform similarly to one another (6,7), and to have reasonable agreement with observed values. The quality of density estimates can be improved by using output feedback, e.g., in (13,14) a mixture-Kalman filter (MKF)-based estimator is applied to two of the SMM modes (FF and CC), and at each time step, the estimator determines which of these modes is stochastically most likely, i.e., whether the FF or CC model is a better fit for each freeway section.…”

K-Means Clustering Method to Classify Freeway Traffic Flow Patterns

“…The model in [7] is simplified in [5] by reducing modes of flow states to two, which are free and congested. A comprehensive extended Kalman filter approach for the estimation and prediction of freeway flow states those are represented by the sectional density is resented in [1] and [6].…”

“…In these studies, the flow state equation has been sought under different approaches. Some successful trials incorporating various approaches include; constructing linear state equation through flow measure based weighting scheme with interpolation [4]; utilizing linear measurement equations to employ flow measures under Eulerian framework [5]; employing filtering methods integrated to high-order traffic flow models [1], [6] using piecewise linear equations of biregime cell based traffic flow model [7]; and establishing linear measure equations to process moving observer data under Lagrangian framework [8].…”