IRUS Total

Unsupervised streaming anomaly detection for instrumented infrastructure

File Description SizeFormat 
AOAS2002-039R2A0.pdfAccepted version4.31 MBAdobe PDFView/Open
Title: Unsupervised streaming anomaly detection for instrumented infrastructure
Authors: Holtegebaum, H
Adams, N
Lau, D-H
Item Type: Journal Article
Abstract: Structural Health Monitoring (SHM) often involves instrumenting structures with distributed sensor networks. These networks typically provide high frequency data describing the spatio-temporal behaviour of the assets. A main objective of SHM is to reason about changes in structures’ behaviour using sensor data. We construct a streaming anomaly detection method for data from a railway bridge instrumented with a fibre-optic sensor network. The data exhibits trend over time, which may be partially attributable to environmental factors, calling for temporally adaptive estimation. Exploiting a latent structure present in the data motivates a quantity of interest for anomaly detection. This quantity is estimated sequentially and adaptively using a new formulation of streaming Principal Component Analysis. Anomaly detection for this quantity is then provided using Conformal Prediction. Like all streaming methods, the pro-posed method has free control parameters which are set using simulations based on bridge data. Experiments demonstrate that this method can operate at the sampling frequency of the data while providing accurate tracking of the target quantity. Further, the anomaly detection is able to detect train passage events. Finally the method reveals a previously unreported cyclic structure present in the data.
Issue Date: 1-Sep-2021
Date of Acceptance: 2-Dec-2020
URI: http://hdl.handle.net/10044/1/86100
DOI: 10.1214/20-AOAS1424
ISSN: 1932-6157
Publisher: Institute of Mathematical Statistics
Start Page: 1101
End Page: 1125
Journal / Book Title: Annals of Applied Statistics
Volume: 15
Issue: 3
Copyright Statement: © 2021 Institute of Mathematical Statistics
Keywords: Science & Technology
Physical Sciences
Statistics & Probability
Structural health monitoring
streaming PCA
stochastic gradient descent
adaptive estimation
conformal prediction
Statistics & Probability
0104 Statistics
1403 Econometrics
Publication Status: Published
Online Publication Date: 2021-09-23
Appears in Collections:Statistics
Faculty of Natural Sciences