Data-driven predictive maintenance : forecasting structural strain in infrastructures using seasonal autoregressive models with exogenous variables

Karami, Niloofar and Poirier, Eric Andrew and Motamedi, Ali; Moreno-Rangel, Alejandro and Kumar, Bimal, eds. (2025) Data-driven predictive maintenance : forecasting structural strain in infrastructures using seasonal autoregressive models with exogenous variables. In: EG-ICE 2025. University of Strathclyde Publishing, GBR. ISBN 9781914241826 (https://doi.org/10.17868/strath.00093293)

Preview

Text. Filename: Karami-etal-EG-ICE-2025-Data-driven-predictive-maintenance-forecasting-structural-strain.pdf Final Published Version License: Download (1MB)| Preview

Abstract

In this study, the effectiveness of Using Seasonal Autoregressive Models with Exogenous Variables (SARIMAX) for predicting strain variations in bridges was investigated by incorporating environmental variables such as temperature and relative humidity. After feature selection and model tuning using ACF, PACF plots, and grid search, the optimal model SARIMAX(1,1,0)(1,0,0,24) was identified. Multiple training window durations, ranging from 3 days to 12 months, were evaluated using expanding window validation techniques. Results indicate that a 5-day training window achieves the lowest average RMSE, demonstrating optimal performance for short-term predictions. Forecast horizon analysis reveals that shorter horizons provide higher predictive accuracy, while longer durations introduce additional uncertainty. Rolling window validation confirmed the model's robustness, maintaining consistent performance across multiple forecast sets. Future work will explore advanced machine learning models to enhance predictive accuracy and robustness.

• Share and Export
  

  ReferData Cite XMLRIOXX2 XMLRDF+N3ASCII CitationEndNoteOpenURL ContextObject in SpanMultiline CSVMODSRefWorksJSONAtomRDF+N-TriplesRIS (EndNote Online, Zotero, Ref manager, etc)EP3 XMLMPEG-21 DIDLRDF+XMLSimple MetadataHTML CitationDublin CoreBibTeXOpenURL ContextObjectMETS
• Item metadata

  Item type:	Book Section
  ID code:	93293
  Dates:	Date Event 1 July 2025 Published 9 June 2025 Accepted
  Subjects:	Fine Arts > Architecture
  Department:	Faculty of Engineering > Architecture
  Depositing user:	Pure Administrator
  Date deposited:	27 Jun 2025 15:15
  Last modified:	01 Jul 2025 11:40
  URI:	https://strathprints.strath.ac.uk/id/eprint/93293