Key performance indicators using a robust prediction modelling to treat squats in railway infrastructures

April 5, 2015 in

Conference Paper


Author
Ali Jamshidi

Co-authors
Alfredo Núñez, Maria Molodova, Zili Li, Rolf Dollevoet

Theme(s)



Conference
Railways 2016 The Third International Conference on Railway Technology: Research, Development and Maintenance
Year: April 2016
Location: Cagliari, Sardinia, Italy

Keywords
ABA measurement, key performance indicators (KPI’s), squat

Link or Download
Click here

Abstract

This paper proposes a growth model for squats in railway infrastructures for design of key performance indicators (KPI’s) in a maintenance time horizon. The main idea of the paper is to make the growth model robust and predictive by the capturing all possible growth scenarios over time. The squats are detected using an axle box acceleration (ABA) measurement system. A methodology is proposed to estimate visual length of squats using the power spectral density of the ABA signal. Next, a robust model is employed for predicting the visual length, including fast, average and slow growth prediction scenarios. The purpose of using the robust model is to consider stochastisities in the squat growth  in order to cover the most important uncertainties. Relying on the prediction model, five KPIs are defined to reflect track condition over time, in 5 different segments of the track Eindhoven-Weert in the Dutch railway network. By using the proposed prediction model, infrastructure manager will be able to plan a condition based maintenance for tracks.