01558131

Component

A methodology for the estimation of queue length at unmetered ramps in real-time has been developed that would assist effective traffic management. The model is based on a Kalman filtering framework that requires input from a supervised machine learning algorithm. Two machine learning algorithms, namely artificial neural networks and k-nearest neighbor, have been investigated. The developed framework utilizes traffic volume counts and occupancy data obtained from loop detectors installed on the ramp. The model has been evaluated using micro-simulation data. The results show that, on average, the model is able to estimate the number of vehicles in the queue within an accuracy of approximately ± 1.3 vehicles for a ramp of length 300 m. A sensitivity analysis of the proposed model was also conducted by changing the number of detectors on the ramp and the reliability of volume counts obtained from the loop detectors.

This paper was sponsored by TRB committee ABJ70 Artificial Intelligence and Advanced Computing Applications.

01550057

15-3468

English

22p

2015

Transportation Research Board 94th Annual Meeting

Digital/other

Figures; References; Tables

Estimating; Highway traffic control; Interchange ramps; Kalman filtering; Machine learning; Neural networks; Real time data processing; Traffic queuing

K-nearest neighbor algorithm; Traffic queue length

Highways; Operations and Traffic Management; I73: Traffic Control

Transportation Research Board Annual Meeting 2015 Paper #15-3468

TRIS, TRB, ATRI

Dec 30 2014 1:10PM