Skip to content

A (prototype) model for prediction of weather-related rail incidents in Great Britain

License

Notifications You must be signed in to change notification settings

mikeqfu/network-rail-weather-rail-incident-prediction

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Understanding and predicting weather-related incidents on GB's rail network

This repo presents data models for predicting weather-related incidents on GB's rail network. Two case study examples are demonstrated in the context of delays caused primarily by high winds and high temperatures, respectively.

GitHub top language GitHub code size in bytes GitHub

Brief background

The prototype data model for the wind-related rail incidents was initially explored in a research project which was intended to establish a modelling framework for establishing the value of existing data resources as supporting data sets for improved decision-making in the UK rail industry. The methodology was firstly demonstrated in the context of delays as a result of high-wind events on the rail network. This case study examined the sub-networks most susceptible to the wind-related delays; and it identified appropriate data sources that could support the decision-making process around placement of trackside weather stations to monitor the impact of the high wind speeds.

The study investigated a variety of data sources, including historical records of rail incidents (attributed to poor weather), and weather observations and vegetation conditions in the vicinity of the incident locations. It showed on how those data could be modelled with logistic regression analysis, based on data integration in both temporal and spatial contexts. A number of factors contributing to the wind- / heat-related incidents have been identified and then used for making predictions of the occurrences of such incidents.

The prototype data models are generic and should also be adaptable to other categories of weather-related incidents, or a wider range of common industry tasks.


Journal article

  • Fu, Q. and Easton, J. M. (2018), Prediction of weather-related incidents on the rail network: prototype data model for wind-related delays in Great Britain. ASCE-ASME Journal of Risk and Uncertainty in Engineering Systems, Part A: Civil Engineering 4(3): 04018027. doi:10.1061/AJRUA6.0000975.

Technical report

  • Fu, Q. and Easton, J. M. (2018), Establishing the value of existing data sources as supporting datasets for improved decision-making. Approved by Network Rail. (Unpublished)

Seminar

  • Fu, Q. Understanding and predicting weather-related incidents on the rail network: case studies of wind-/heat-related incidents. Invited seminar at the Institute for Risk and Uncertainty, University of Liverpool, Liverpool, UK, 8 January 2019.

Conference papers

Open-source Python packages