Abstract

Vibration-based damage detection in civil structures using data-driven methods requires sufficient vibration responses acquired with a sensor network. Due to technical and economic reasons, it is not always possible to deploy a large number of sensors. This limitation may lead to partial information being handled for damage detection purposes, under environmental variability. To address this challenge, this article proposes an innovative multi-level machine learning method by employing the autoregressive spectrum as the main damage-sensitive feature. The proposed method consists of three levels: (i) distance calculation by the log-spectral distance, to increase damage detectability and generate distance-based training and test samples; (ii) feature normalization by an improved factor analysis, to remove environmental variations; and (iii) decision-making for damage localization by means of the Jensen–Shannon divergence. The major contributions of this research are represented by the development of the aforementioned multi-level machine learning method, and by the proposal of the new factor analysis for feature normalization. Limited vibration datasets relevant to a truss structure and consisting of acceleration time histories induced by shaker excitation in a passive system, have been used to validate the proposed method and to compare it with alternate, state-of-the-art strategies.

Highlights

  • Civil structures must be monitored to detect, ideally in real time, any damage due to aging, material deterioration or unexpectedly large excitations

  • In order to deal with the aforementioned limitations and challenges, this article proposes a parametric spectral-based feature extraction approach and an innovative multi-level machine learning method for early damage detection in cases characterized by a limited number of sensors and under environmental variability

  • Having considered the multivariate datasets Ex and Ez provided by feature normalization, we propose to use the JS-divergence method with the aid of kernel density estimation (KDE) for early damage detection

Read more

Summary

Introduction

Civil structures must be monitored to detect, ideally in real time, any damage due to aging, material deterioration or unexpectedly large excitations. Despite the success and applicability of various feature extraction and statistical analysis techniques, the adoption of a limited number of sensors may prevent these approaches from capturing sufficient dynamic characteristics or damage-sensitive features to provide reliable damage detectability This issue becomes even worse when the limited information is coupled with the environmental and/or operational variability conditions. In order to deal with the aforementioned limitations and challenges, this article proposes a parametric spectral-based feature extraction approach and an innovative multi-level machine learning method for early damage detection in cases characterized by a limited number of sensors and under environmental variability. It has been observed that the proposed multi-level machine learning method is able to accurately detect damage in cases characterized by a limited sensor deployment and environmental variability, owing essentially to the improved factor analysis approach

Parametric Spectral-Based Feature Extraction by AR Modeling
Proposed Multi-Level Machine Learning Method
Level I
Level II
Classical Factor Analysis
Markov Chain Monte Carlo Factor Analysis
Determination of the Number of Factors
Level III
Relative Entropy Measures in Information Theory
Damage Detection Scheme
Case Study
Response Modeling and Feature Extraction
Comparative Studies
Findings
Conclusions
Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call