John Mc Keon
MPhil Researcher
Email: C02876523@mytudublin.ie
Project Title: An Analysis of the Effect of High Friction Surfacing on Accident Rates Using Open Data, Network Analysis and Machine Learning.
Supervisor: Dr. Santos Fernandez Noguerol
The main aim of this project is to test the effectiveness of High Friction Surfacing (HFS) at reducing accident rates. HFS is a layer of granular material held in place by a binder to the road surface. The purpose of the layer is to increase the friction between vehicles and the road surface.
This project looks to use a novel approach to studying accident rates. As flow is a key indicator for traffic accidents this project is looking to estimate traffic flow at each junction in Dublin’s road network. We will create a network model with the nodes as junctions and the edges as roads. With this graph we can use centrality formula to calculate the flow of traffic. We will use Dublin’s intelligent transport system (SCATS) to calibrate the model. This will give an estimate of traffic flow on primary, secondary and tertiary roads.
Calculating centrality on a network graph is computationally intensive. To make the model more efficient we will use a Long Short-Term (LSTM) neural network. A neural network is a computer algorithm that looks to find relationships in a data set. An LSTM model is noted for being the most efficient model for time series data sets.
With the traffic flow data calculated a regression analysis will be conducted on junctions with HFS and junctions without. In this analysis we will control for traffic flow rate. The analysis should give a highly accurate value for the effectiveness of HFS.