View in EDS

Enhancing structural health monitoring through linearity feature analysis and marginal distribution-based probabilistic models.

Saved in:
Bibliographic Details
Title: Enhancing structural health monitoring through linearity feature analysis and marginal distribution-based probabilistic models.
Authors: Puerto-Santana, Cristian1,2 (AUTHOR) cristian.puerto@upc.edu, Ocampo-Martinez, Carlos2 (AUTHOR) carlos.ocampo@upc.edu, Diaz-Rozo, Javier1 (AUTHOR) jdiaz@ainguraiiot.com, Enqvist, Per3 (AUTHOR) penqvist@math.kth.se
Source: Engineering Structures. Aug2026, Vol. 360, pN.PAG-N.PAG. 1p.
Subjects: Structural health monitoring, Copula functions, System identification, Outlier detection, Mechanical vibration research, Feature extraction, Statistical models
Abstract: Infrastructure engineering is a critical discipline focused on the design, development, and maintenance of structures essential to society. The necessity for ongoing maintenance has led to the development of standards and regulations aimed at ensuring structural safety and durability. Structural Health Monitoring (SHM) addresses this need by using sensors and intelligent systems to detect anomalies. Recent efforts have introduced probabilistic modeling to improve SHM practices. This study addresses challenges in feature extraction and modeling by proposing a multivariate, data-driven methodology. The proposed method combines the Hankel alternative view of Koopman (HAVOK) system identification framework for generating features with a Copula-based probabilistic model and for computing a condition indicator to detect structural anomalies. This method was validated using one year of vibration data from four accelerometers installed on the "Andoian DF 12" bridge in Guipuzkoa, Spain. Results show that the proposed framework improves the condition indicator by reducing outliers and variance while enhancing model robustness. Comparisons with other probabilistic models confirm the proposed approach's efficiency in both accuracy and computational performance up to 10% more in line with the training data regarding conventional probabilistic models and a training time that is up to 11 times faster than variational auto encoders models for the tested conditions. • A method for monitor structures based on free response estimations is proposed. • A method using marginal feature distribution for robust anomaly detection is presented. • This method generates a robust condition indicator for structural health monitoring. • One year of vibration data from a real bridge validates the proposed approach. • The method is experimentally compared with other approaches by time and resource usage. [ABSTRACT FROM AUTHOR]
Copyright of Engineering Structures is the property of Elsevier B.V. and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)
Database: Engineering Source
Description
Abstract:Infrastructure engineering is a critical discipline focused on the design, development, and maintenance of structures essential to society. The necessity for ongoing maintenance has led to the development of standards and regulations aimed at ensuring structural safety and durability. Structural Health Monitoring (SHM) addresses this need by using sensors and intelligent systems to detect anomalies. Recent efforts have introduced probabilistic modeling to improve SHM practices. This study addresses challenges in feature extraction and modeling by proposing a multivariate, data-driven methodology. The proposed method combines the Hankel alternative view of Koopman (HAVOK) system identification framework for generating features with a Copula-based probabilistic model and for computing a condition indicator to detect structural anomalies. This method was validated using one year of vibration data from four accelerometers installed on the "Andoian DF 12" bridge in Guipuzkoa, Spain. Results show that the proposed framework improves the condition indicator by reducing outliers and variance while enhancing model robustness. Comparisons with other probabilistic models confirm the proposed approach's efficiency in both accuracy and computational performance up to 10% more in line with the training data regarding conventional probabilistic models and a training time that is up to 11 times faster than variational auto encoders models for the tested conditions. • A method for monitor structures based on free response estimations is proposed. • A method using marginal feature distribution for robust anomaly detection is presented. • This method generates a robust condition indicator for structural health monitoring. • One year of vibration data from a real bridge validates the proposed approach. • The method is experimentally compared with other approaches by time and resource usage. [ABSTRACT FROM AUTHOR]
ISSN:01410296
DOI:10.1016/j.engstruct.2026.122754