Damage detection and the classification of carbon fiber-reinforced composites using non-destructive testing (NDT) techniques are of great importance. This paper applies an acoustic emission (AE) technique to obtain AE data from three tensile damage tests determining fiber breakage, matrix cracking, and delamination. This article proposes a deep learning approach that combines a state-of-the-art deep learning technique for time series classification: the InceptionTime model with acoustic emission data for damage classification in composite materials. Raw AE time series and frequency-domain sequence data are used as the input for the InceptionTime network, and both obtain very high classification performances, achieving high accuracy scores of about 99%. The InceptionTime network produces better training, validation, and test accuracy with the raw AE time series data than it does with the frequency-domain sequence data. Simultaneously, the InceptionTime model network shows its potential in dealing with data imbalances.

This study investigated the mechanism of delamination damage in the double cantilever beam (DCB) standard test by the use of the strain energy release rate. The curve of the strain energy release rate was verified by the Rise Angle (RA) method. For this purpose, 24-layer carbon fiber/epoxy multidirectional laminates with interface orientations of 0°, 30°, 45°, and 60° were fabricated according to the standard ASTM D5528(13). In the course of this test, acoustic emission (AE) was used for real-time monitoring, and combined with micro visualization, the damage mechanism of composite multidirectional laminates was studied at multiple scales. Combining the AE detection results with micro visualization, it is found that the AE parameters and the damage to multidirectional laminates could realize a one-to-one correspondence. Through the study of the variation of the RA value, load, and strain energy release rate with the crack length, it is proved that the AE parameters can effectively characterize the initiation of delamination damage.

Two mesomechanics models were analyzed in an attempt to reveal the relationship between stored energy and back stress. It has been indicated that the portion of elastic stored energy due to residual microstresses (E_SR) is closely related to intergranular back stress (X_inter), and the stored energy of dislocations inside grains (E_SD) can be estimated with the plastic work of intragranular back stress (X_intra). Then, the evolution of back stress during cyclic loading was studied, and the plastic work of back stress (W_pB) was calculated with the low cycle fatigue experimental data of Ti-6Al-4V. The result shows that W_pB is partially released at every reverse loading, sufficient to reproduce the evolution of stored energy correctly under cyclic loading. The study also reveals that partially released energy is related to the decrease of X_inter at the initial state of reversal loading resulting from the reduction of the plastic strain incompatibility between grains.

This study resolves the long-standing question of the origin of bilinear Low Cycle Fatigue (LCF) behavior in Ti-6Al-4V using a high-fidelity CPFEM-XFEM framework. We identify that the fundamental origin lies in a fundamental shift in the efficiency of converting macroscopic energy dissipation into microscopic damage. This energetic efficiency is directly governed by the evolution of plastic strain heterogeneity (quantified by the Coefficient of Variation, CV). At low strain amplitudes, high strain localization (high CV) creates a highly efficient “energy funnel,” concentrating dissipated energy into a few critical grains. This manifests physically as a single-crack failure mode, where the crack initiation phase is prolonged, consuming ~80% of the total fatigue life. Conversely, at high strain amplitudes, deformation homogenization (low CV) leads to inefficient, diffuse energy dissipation across many grains. The material must therefore activate a more drastic failure mechanism—multi-site crack initiation and coalescence—to accumulate sufficient damage, reducing the initiation phase to just ~45% of the total life. Therefore, the bilinear C-M curve is the macroscopic signature of this transition from an energetically efficient, localized damage mode to an inefficient, distributed one. This work provides a quantitative, mechanism-based framework for understanding and predicting the complex fatigue behavior of advanced metallic materials.

—Glass fiber-reinforced composites (GFRP) are prone to undetectable damage, posing challenges to their safe application in aerospace, transportation, and other fields. This study integrated acoustic emission (AE), microscopic visualization, and digital image correlation (DIC) techniques to monitor and quantify damage in multidirectional pre-perforated GFRP laminates under tensile-tensile fatigue loading. It analyzed the intrinsic correlations between surface strain evolution, AE characteristic parameters (amplitude, energy, duration, etc.), and damage modes (matrix cracking, delamination, fiber breakage) of laminates with different fiber orientations, and established the AE-based damage variable D_AE. Its validity was verified using stress-strain curves and damage micromorphology. Results showed that fiber orientation significantly affected the mechanical response and damage behavior of laminates. AE parameters and surface strain effectively characterized damage mechanisms, with obvious differences in strain distribution, stress-strain curves, and AE signals among laminates with different orientations. The engineering error of D_AE for fatigue damage characterization was within 20%, confirming the feasibility of AE technology in quantifying composite fatigue damage and providing key support for composite structural damage monitoring and life prediction.