الفهرس 
Title : Development Of Operating Speed Models For Multilane Arterial Roads Using Artificial Neural Networks
SummaryOnly 14 pages are availabe for public view
 |  | from | 117 |  |  |
| | | from | 117 | | |
Abstract
The operating speed of vehicles on a road under free-flow conditions is considered one of the important parameters that evaluates the trip quality. Operating speed models help assess and evaluate the speed changes along successive road elements. Reducing speed changes along the road alignment improves the traffic performance and increases safety. Although several geometric elements of the road affect operating speeds, including horizontal curve radii, cross-section elements, and vertical grades; most operating speed models using regression techniques capture mainly the impact of horizontal curve elements. In addition, most models are developed successfully on two-lane two-way highways where speeds are affected more by the variation of the geometric features. This research investigates the use of Artificial Neural Networks (ANN) as a tool to develop operating speed models for multilane urban elevated arterial roads. Parameters investigated in the study of operating speeds included geometric features of the element and the residual impact of upstream elements. A data collection exercise was undertaken on two major elevated arterial roads in Greater Cairo Region GCR, Egypt; namely, Saft Al-Laban and 6th of October corridors. Each travel corridor was divided into several road segments with homogeneous geometric features (number of lanes and curvature). Speed data was extracted from google API distance matrix and validated using test vehicle speed data. A regression-based modelling exercise was undertaken in the preliminary investigation phase to serve as a benchmark for the intended machine learning modelling exercise. Several ANN architectures -with different combinations of input variables- were developed and evaluated. The residual effect of operating speeds on upstream road segments was investigated through the incorporation of upstream segments’ speeds as input variables. ANN testing results revealed the ability of the developed models to predict operating speeds. Results showed that the prediction power of the developed ANN models – capturing the residual effect of upstream speeds- outperformed regression-based ones.