Enhancing generalizability of a machine learning model for infrared thermographic defect detection - using 3D numerical modeling

Aug 6, 2026

Infrared (IR) thermography enables non-destructive testing by analyzing thermal patterns to reveal structural defects and thermal anomalies, but its performance can be limited by heat diffusion, environmental sensitivity, and operator constraints. This work studies how to improve the generalizability of machine learning models for IR thermographic defect detection using 3D numerical modeling and systematic parameter variation. Multiple test datasets with unseen parameter changes are used to evaluate robustness, identify sensitivity to defect depth, thermal conductivity, and specimen thickness, and guide the design of balanced training data.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages15

Reading Time38 min

AccessFree to download

Introduction

Background of IR thermographic NDT

Flash thermography and machine learning integration

Problem of generalizability

Limits under fixed training setups

Study approach

3D numerical modeling and parameter variability

Evaluation with multiple unseen test datasets

A. Chulkov et alii, Frattura edIntegrità Strutturale, 70 (2024) 177-191; DOI: 10.3221/IGF-ESIS.70.10 | Enhancing generalizability of a machine learning model for infrared thermographic defect detection by using 3d numerical modeling
Arsenii Chulkov
Tomsk Polytechnic University, Russia
[chulkovao@tpu.ru](chulkovao@tpu.ru), [http://orcid.org/0000-0003-1226-0013](http://orcid.org/0000-0003-1226-0013)
Alexey Moskovchenko
University of West Bohemia, Czech Republic [alexeym@ntc.zcu.cz](alexeym@ntc.zcu.cz), [http://orcid.org/0000-0002-2813-2529](http://orcid.org/0000-0002-2813-2529)
Vladimir Vavilov
Tomsk Polytechnic University, Russia
[vavilov@tpu.ru](vavilov@tpu.ru), [http://orcid.org/ 0000-0002-9828-7374](http://orcid.org/ 0000-0002-9828-7374) | | | --- | --- | | |
Citation: Chulkov, A., Moskovchenko, A., Vavilov, V., Enhancing generalizability of a machine learning model for infrared thermographic defect detection by using 3D numerical modeling, Frattura ed Integrità Strutturale, 70 (2024) 177-191.
Received: 28.05.2024
Accepted: 26.07.2024
Published: 19.08.2024
Issue: 10.2024
Copyright: © 2024 This is an open access article under the terms of the CC-BY 4.0, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. | | KEYWORDS. Infrared thermography, Nondestructive Testing, Machine learning, Numerical simulation, Defect detection. | | | INTRODUCTION
I nfrared (IR) thermography is a method of non-destructive testing (NDT) based on the analysis of thermal patterns on
the surface of objects under test by using thermal imagers [1]. Thermal stimulation of objects and subsequent analysis of temperature distributions allow detecting structural defects and thermal anomalies in various materials. Due to its simplicity, non-contact nature of testing and capacity to swiftly assess large areas, IR thermographic NDT has become | | A. Chulkov et alii, Frattura edIntegrità Strutturale, 70 (2024) 177-191; DOI: 10.3221/IGF-ESIS.70.10 widespread in the aerospace industry [2, 3], power production and civil engineering [4–6], including evaluation of composite materials (carbon and glass fiber plastics) [7] . However, IR thermography possesses some drawbacks, such as a diffusive nature of heat conduction, sensitivity to environmental conditions, need for properly-trained operators, etc. Flash thermography (FT) involves the use of powerful heat sources (e.g. flash tubes and lasers) generating short optical pulses which can be considered as Dirac pulses [8] . Due to short observation times, flash thermography provides highresolution thermal images facilitating the detection of small-size defects in materials and structures. The integration of automation and machine learning into IR thermography has significantly expanded potentials of this technique [9–11] . In 2016, Khodayar et al. outlined the use of artificial intelligence as the “2050-horizon” in IR thermographic NDT [12] . The state-of-the-art and recent improvements in Convolutional Neural Networks (CNN) was presented in 2018 by Jiuxiang Gu et al. [13] . In the recent review paper, Yunze He et al. stated that the rapid development of deep learning makes IR machine vision and IR thermographic NDT more intelligent thus contributing to broadening applications of these techniques [14] . In active IR thermographic NDT of materials, the principles of deep learning can be helpful in solving inverse heat conduction problems, i.e. performing defect characterization of hidden defects that is a permanent challenge in thermal NDT (TNDT) . For example, Yousefi et al. showed that CNNs can serve as an unsupervised extractor of defect features in IR NDT [15]. Haiyi Wu et al. proposed the deep learning model based on CNNs with a U-shape architecture to predict the heterogeneous distribution of circle-shaped fillers in composites [16] . In pulsed TNDT, a couple of different CNNs

FAQs

Why is improving generalizability important for ML-based infrared thermographic defect detection?

Neural network models often perform well only under specific training conditions and datasets. When parameters differ from the training setup, defect detection and characterization capabilities can degrade.

How does this study evaluate generalizability?

It performs comprehensive evaluation by systematically varying numerical model parameters and testing trained models on multiple test datasets containing unseen parameter variations.

Which parameters are varied to understand their influence on model performance?

The study manipulates defect depth, material thermal conductivity, and sample thickness to assess how these factors affect detection performance and robustness.

Aug 6, 2026

Upload by Emma Wilson