This paper proposes a methodology for generating virtual samples of stiffness degradation of composite laminates to facilitate the training of artificial neural network (ANN)-based predictive models. Considering the impact of matrix cracks, delamination, and fiber fracture on the stiffness of laminated panels, a finite element model driven by the physical mechanism of fatigue damage is established for generating virtual samples of fatigue evolution. The stiffness and evolution of virtual samples for [0/903/0/903]S glass fiber reinforced polymer (GFRP) laminates are obtained. By employing the fatigue evolution prediction method based on β-variational autoencoder (β-VAE) and neural ordinary differential equation networks (Neural ODE), the proposed approach effectively learns the underlying mechanisms in fatigue evolution data and shows good correlation to experimental observation. The results indicate that physics-driven virtual sample generation significantly enhances the predictive accuracy and robustness of neural network models, broadening their practical applicability.
WanhillRMolentLBarterS, et al.Milestone case histories in aircraft structural integrity, 2015.
2.
GamstedtETalrejaR. Fatigue damage mechanisms in unidirectional carbon-fibre-reinforced plastics. J Mater Sci1999; 34: 2535–2546. https://doi.org/10.1023/a:1004684228765
3.
HighsmithALReifsniderKL. Stiffness-reduction mechanisms in composite laminates. Damage in composite materials: basic mechanisms, accumulation, tolerance, and characterization. ASTM International, 1982.
4.
TalrejaRSinghCV. Damage and failure of composite materials. Cambridge University Press, 2012.
5.
TaoCCJiHLQiuJH, et al.Characterization of fatigue damages in composite laminates using Lamb wave velocity and prediction of residual life. Compos Struct2017; 166: 219–228. https://doi.org/10.1016/j.compstruct.2017.01.034
6.
ShiriSYazdaniMPourgol-MohammadM. A fatigue damage accumulation model based on stiffness degradation of composite materials. Mater Des2015; 88: 1290–1295. https://doi.org/10.1016/j.matdes.2015.09.114
7.
Van PaepegemWDegrieckJ. A new coupled approach of residual stiffness and strength for fatigue of fibre-reinforced composites. Int J Fatig2002; 24(7): 747–762. https://doi.org/10.1016/s0142-1123(01)00194-3
SendeckyjGP. Life prediction for resin-matrix composite materials. Composite materials series, Elsevier1991, pp. 431–483.
14.
KassapoglouC. Fatigue model for composites based on the cycle-by-cycle probability of failure: implications and applications. J Compos Mater2011; 45(3): 261–277. https://doi.org/10.1177/0021998308104357
15.
ChengZ-QTanWXiongJ-J. Progressive damage modelling and fatigue life prediction of plain-weave composite laminates with low-velocity impact damage. Compos Struct2021; 273: 114262. https://doi.org/10.1016/j.compstruct.2021.114262
16.
IarveEVHoosKBraginskyM, et al.Progressive failure simulation in laminated composites under fatigue loading by using discrete damage modeling. J Compos Mater2017; 51(15): 2143–2161. https://doi.org/10.1177/0021998316681831
17.
NasserJGrooLSodanoH. Artificial neural networks and phenomenological degradation models for fatigue damage tracking and life prediction in laser induced graphene interlayered fiberglass composites. Smart Mater Struct2021; 30(8): 085010. https://doi.org/10.1088/1361-665x/ac093d
18.
FengSHanXMaZ, et al.Data-driven algorithm for real-time fatigue life prediction of structures with stochastic parameters. Comput Methods Appl Mech Eng2020; 372: 113373. https://doi.org/10.1016/j.cma.2020.113373
19.
YangJYKangGZKanQH. A novel deep learning approach of multiaxial fatigue life-prediction with a self-attention mechanism characterizing the effects of loading history and varying temperature. Int J Fatig2022; 162: 106851. https://doi.org/10.1016/j.ijfatigue.2022.106851
20.
TaoCCZhangCJiHL, et al.Application of neural network to model stiffness degradation for composite laminates under cyclic loadings. Compos Sci Technol2021; 203: 108573. https://doi.org/10.1016/j.compscitech.2020.108573
21.
TaoCCZhangCJiHL, et al.Fatigue damage characterization for composite laminates using deep learning and laser ultrasonic. Compos Part B-Eng2021; 216: 108816. https://doi.org/10.1016/j.compositesb.2021.108816
22.
JiangDQianHWangY, et al.Data driven prediction of fatigue residual stiffness of braided ceramic matrix composites based on Latent-ODE. Compos Struct2023; 323: 117504. https://doi.org/10.1016/j.compstruct.2023.117504
23.
ChenZGZhengTZhangL, et al.A deep learning unified model for predicting the residual stiffness of plain weave composites under combined high and low cycle fatigue loading. Int J Fatig2025; 199: 109035. https://doi.org/10.1016/j.ijfatigue.2025.109035
24.
HuGBTaoCCZhangC, et al.Investigation on the effect of ply orientations on the fatigue behavior of composite laminates: from the perspective of deep learning. Compos Part B-Eng2026; 313: 113408. https://doi.org/10.1016/j.compositesb.2026.113408
25.
LeeLFuKYangJ. Prediction of fatigue damage and life for composite laminates under service loading spectra. Compos Sci Technol1996; 56(6): 635–648. https://doi.org/10.1016/0266-3538(96)00043-7
26.
KashtalyanMSoutisC. The effect of delaminations induced by transverse cracks and splits on stiffness properties of composite laminates. Compos Part A-Appl S2000; 31(2): 107–119. https://doi.org/10.1016/s1359-835x(99)00066-4
27.
ShenHJYaoWXQiWX, et al.Experimental investigation on damage evolution in cross-ply laminates subjected to quasi-static and fatigue loading. Compos Part B-Eng2017; 120: 10–26. https://doi.org/10.1016/j.compositesb.2017.02.033
28.
WilsonDL. The analysis of survival (mortality) data - fitting gompertz, weibull, and logistic functions. Mech Ageing Dev1994; 74(1-2): 15–33. https://doi.org/10.1016/0047-6374(94)90095-7
29.
JohnsonDSargeantAAllenS. Fitting Richards' curve to data of diverse origins. Growth1975; 39(3): 315–330.
YokozekiTAokiTOgasawaraT, et al.Effects of layup angle and ply thickness on matrix crack interaction in contiguous plies of composite laminates. Compos Appl Sci Manuf2005; 36(9): 1229–1235. https://doi.org/10.1016/j.compositesa.2005.02.002
32.
MoureMGarcía-CastilloSSánchez-SáezS, et al.Influence of ply cluster thickness and location on matrix cracking evolution in open-hole composite laminates. Compos B Eng2016; 95: 40–47. https://doi.org/10.1016/j.compositesb.2016.03.091
HuangYZhangCTaoC, et al.Fatigue evolution prediction for fiber‐reinforced plastics based on frequency‐wavenumber wavefield of guided wave using deep‐learning model. Fatigue Fract Eng M2024; 47(2): 549–564. https://doi.org/10.1111/ffe.14192
35.
RubanovaYChenRTDuvenaudDK. Latent ordinary differential equations for irregularly-sampled time series. Adv Neural Inf Process Syst2019; 32: ■■■.
36.
ChenRTRubanovaYBettencourtJ, et al.Neural ordinary differential equations. Adv Neural Inf Process Syst2018; 31: ■■■.
37.
BurgessCPHigginsIPalA, et al.Understanding disentangling in $/beta $-VAE, 2018. arXiv preprint arXiv:1804.03599.
38.
HigginsIMattheyLPalA, et al.beta-vae: learning basic visual concepts with a constrained variational framework. ICLR (Poster)2017; 3: ■■■.
