This paper presents an incremental transfer learning (TL) approach for monitoring fatigue crack sizes in welded joints based on strain variations. The proposed method integrates a pretrained machine learning (ML) model, developed from a calibrated numerical database, with incremental TL via limited experimental data to enhance the accuracy of the crack size estimation. Compared with experimental measurements, the pretrained ML model shows escalating errors in the crack profile estimations as the fatigue crack evolves, while the proposed incremental TL approach provides an effective solution to mitigate this issue. Validation on one bending and three tensile cruciform welded specimens demonstrates the strong predictive performance of the proposed incremental TL approach. The study also reveals the effect of fine-tuning frequency on the accuracy of crack size prediction, providing practical guidance for model updating strategies during fatigue crack monitoring. For a practical scenario, where the strain is only available after the crack detection and the strain variation is referenced to the strain measured from a cracked joint, the proposed approach demonstrates accurate estimations of the subsequent crack growth. The success of this method underscores the high potential of cost-effective strain sensors in supporting crack diagnosis and structural health monitoring.
KolsteinM.Fatigue classification of welded joints in orthotropic steel bridge decks. PhD Thesis, Delft University of Technology, Delft, 2007.
2.
ConnorRFisherJGattiW, et al. Manual for design, construction, and maintenance of orthotropic steel deck bridges. Federal Highway Administration, U.S. Department of Transportation, United States, 2012.
3.
YeterBGarbatovYSoaresCG.Fatigue damage assessment of fixed offshore wind turbine tripod support structures. Eng Struct2015; 101: 518–528.
4.
IgwemezieVMehmanparastAKoliosA.Current trend in offshore wind energy sector and material requirements for fatigue resistance improvement in large wind turbine support structures–A review. Renew Sust Energ Rev2019; 101: 181–196.
DuanLWangCZhaiM, et al. Monitoring and evaluation of fatigue damage for orthotropic steel deck using acoustic emission technology. J Traffic Transp Eng2020; 20(1): 60–73.
7.
FengLQianX.Enhanced sizing for surface cracks in welded tubular joints using ultrasonic phased array and image processing. NDT E Int2020; 116: 102334.
8.
YangHWangPQianH, et al. A study of fatigue crack propagation paths at U-rib welds in orthotropic bridge decks using a phased-array imaging technique. Theor Appl Fract Mec2022; 119: 103310.
9.
ZhangDCuiCZhangX, et al. Monitoring fatigue cracks in rib-to-deck joints of orthotropic steel deck using ultrasonic Lamb waves. Thin-Walled Struct2023; 189: 110922.
10.
HoMEl-BorgiSPatilD, et al. Inspection and monitoring systems subsea pipelines: a review paper. Struct Health Monit2020; 19(2): 606–645.
11.
FengLQianX.Size effect and life estimation for welded plate joints under low cycle actions at room and low ambient temperatures. Thin-Walled Struct2018; 132: 195–207.
12.
DungC VSekiyaHHiranoS, et al. A vision-based method for crack detection in gusset plate welded joints of steel bridges using deep convolutional neural networks. Automat Constr2019; 102: 217–229.
13.
KongXLiJ.Vision-based fatigue crack detection of steel structures using video feature tracking. Comput-Aided Civ Inf2018; 33(9): 783–799.
14.
FuZJiBZhangC, et al. Fatigue performance of roof and U-rib weld of orthotropic steel bridge deck with different penetration rates. J Bridge Eng2017; 22(6): 04017016.
15.
LimH JSohnHKimY.Data-driven fatigue crack quantification and prognosis using nonlinear ultrasonic modulation. Mech Syst Signal Pr2018; 109: 185–195.
16.
LiJ.Research on formation mechanism and performance evaluation of dominant fatigue failure modes of orthotropic steel deck. PhD Thesis, Southwest Jiaotong University, Chengdu, China, 2022.
17.
CadiniFSbarufattiCCorbettaM, et al. Particle filtering-based adaptive training of neural networks for real-time structural damage diagnosis and prognosis. Struct Control Hlth2019; 26(12): e2451.
18.
CristianiDSbarufattiCCadiniF, et al. Fatigue damage diagnosis and prognosis of an aeronautical structure based on surrogate modelling and particle filter. Struct Health Monit2021; 20(5): 2726–2746.
19.
MunirNKimHJSongSJ, et al. Investigation of deep neural network with drop out for ultrasonic flaw classification in weldments. J Mech Sci and Technol2018; 32: 3073–3080.
20.
ZhaoMZhouWHuangY, et al. Imaging cracks in orthotropic steel decks based on guided wave and variational Bayesian robust principal component analysis. Struct Health Monit2025: 14759217251320044.
21.
NiuSSrivastavaV.Ultrasound classification of interacting flaws using finite element simulations and convolutional neural network. Eng Comput-Germany2022; 38(5): 4653–4662.
22.
CheokEWWQianXKapsA, et al. A strain-interfaced digital twin solution for corner fatigue crack growth using Bayesian inference. Int J Fatigue2025; 191: 108705.
23.
HuFGouHYangH, et al. Digital twin-based fatigue life assessment of orthotropic steel bridge decks using inspection robot and deep learning. Automat Constr2025; 172: 106022.
24.
GibsonSJRogersTJCrossEJ.Distributions of fatigue damage from data-driven strain prediction using Gaussian process regression. Struct Health Monit2023; 22(5): 3065–3076.
25.
XuHLiuLXuJ, et al. Deep learning enables nonlinear Lamb waves for precise location of fatigue crack. Struct Health Monit2024; 23(1): 77–93.
26.
FengCXuLZhaoL, et al. A state-of-art review on prediction model for fatigue performance of welded joints via data-driven method. Adv Eng Mater2023; 25(9): 2201430.
27.
LiDWangYYanWJ, et al. Acoustic emission wave classification for rail crack monitoring based on synchro squeezed wavelet transform and multi-branch convolutional neural network. Struct Health Monit2021; 20(4): 1563–1582.
28.
ZhangJZhuJGuoW, et al. A machine learning-based approach to predict the fatigue life of three-dimensional cracked specimens. Int J Fatigue2022; 159: 106808.
29.
MaXHeXTuZC.Prediction of fatigue–crack growth with neural network-based increment learning scheme. Eng Fract Mech2021; 241: 107402.
30.
SampathSJangJSohnH.Ultrasonic Lamb wave mixing based fatigue crack detection using a deep learning model and higher-order spectral analysis. Int J Fatigue2022; 163: 107028.
31.
FengLQianX.An adaptive learning approach to determine and update crack sizes from strain relaxation data for welded plate joints. Eng Fract Mech2022; 259: 108165.
32.
FengLQianX.Enhanced real-time crack monitoring and updating in welded structural components with limited measurement data. Mech Syst and Signal Pr2023; 198: 110403.
33.
CheokEWWQianXChenC, et al. A local digital twin approach for identifying, locating and sizing cracks in CHS X-joints subjected to brace axial loading. Eng Struct2024; 299: 117085.
34.
TengSChenXChenG, et al. Structural damage detection based on transfer learning strategy using digital twins of bridges. Mech Syst Signal Pr2023; 191: 110160.
35.
FangXLiuGWangH, et al. Fatigue crack growth prediction method based on machine learning model correction. Ocean Eng2022; 266: 112996.
36.
KatsigiannisSSeyedzadehSAgapiouA, et al. Deep learning for crack detection on masonry façades using limited data and transfer learning. J Build Eng2023; 76: 107105.
37.
ZhangMGuoTZhangG, et al. Missing monitoring data reconstruction for cable-stayed bridge using knowledge transfer-based generative pre-trained model. Struct Health Monit2025: 14759217251324156.
38.
YanoMOFigueiredoEda SilvaS, et al. Foundations and applicability of transfer learning for structural health monitoring of bridges. Mech Syst and Signal Pr2023; 204: 110766.
39.
SongLCaoZSunH, et al. Transfer learning for structure damage detection of bridges through dynamic distribution adaptation. Structures2024; 67: 106972.
40.
PanQBaoYLiH.Transfer learning-based data anomaly detection for structural health monitoring. Struct Health Monit2023; 22(5): 3077–3091.
41.
LiZZhangSLiH, et al. On-line transfer learning for multi-fidelity data fusion with ensemble of deep neural networks. Adv Eng Inform2022; 53: 101689.
42.
PengYZhangYZhangL, et al. Prediction of corrosion fatigue crack growth rate in aluminum alloys based on incremental learning strategy. Int J Fatigue2024; 187: 108481.
43.
RolnickDAhujaASchwarzJ, et al. Experience replay for continual learning. Adv Neural Inform Proc Syst2019; 32: 350–360.
44.
FengLQianX.Low cycle fatigue test and enhanced lifetime estimation of high-strength steel S550 under different strain ratios. Mar Struct2018; 61: 343–360.
45.
ZhangQWeiCXuZ, et al. Experimental study on fatigue performance of integral welded joints in steel truss bridges. China J Highway Transp2022; 35(12): 77–90.
46.
GB/T 714-2015. Structural steel for bridge. General administration of quality supervision, inspection and quarantine of the People’s Republic of China (AQSIQ) and standardization administration of the People’s Republic of China (SAC), 2015.
47.
WangDXiangCMaY, et al. Experimental study on the root-deck fatigue crack on orthotropic steel decks. Mate Design2021; 203: 109601.
48.
LiaoXWeiHPanH, et al. Fatigue resistance of rib-to-bimetallic steel deck welded joints in orthotropic steel bridge decks. Thin-Walled Struct2024; 194: 111318.
49.
Dassault Systèmes Simulia Corp, Abaqus/CAE User’s Manual, Version 2019; 2019.
50.
BaikBYamadaKIshikawaT.Fatigue crack propagation analysis for welded joint subjected to bending. Int J Fatigue2011; 33(5): 746–758.
51.
ZongLShiGWangY Q, et al. Experimental and numerical investigation on fatigue performance of non-load-carrying fillet welded joints. J Constr Steel Res2017; 130: 193–201.
52.
BS7910 B.S. Guide to methods for assessing the acceptability of flaws in metallic structures. British Standards Institution, 2013.
53.
MathWorks, MATLAB, R2020b Documentation; 2020.
54.
HuangMZhangJLiJ, et al. Damage identification of steel bridge based on data augmentation and adaptive optimization neural network. Struct Health Monit2024: 14759217241255042.
55.
ChenDLiYLiuK, et al. A physics-informed neural network approach to fatigue life prediction using small quantity of samples. Int J Fatigue2023; 166: 107270.
56.
LiXFuZShuJ, et al. A modified physics-informed neural network to fatigue life prediction of deck-rib double-side welded joints. Int J Fatigue2024; 189: 108566.
57.
MartinDKühlNSatzgerG.Virtual sensors. Bus Inform Syst Eng2021; 63: 315–323.
