Aiming at the problems of complex dispersion, insufficient feature representation and poor interpretability of deep learning models in Lamb wave signal processing for composite materials, a Lamb wave damage identification method based on time-frequency dual-domain feature fusion and interpretable deep learning is proposed to improve the accuracy and reliability of damage identification. Time domain and frequency domain features of Lamb wave signals are extracted to construct a dual-domain feature system with time domain features as the core, and a multi-branch 1D-CNN is designed for targeted learning and efficient fusion of the dual-domain features. A systematic data augmentation strategy is introduced to enhance the model's generalization ability under small sample conditions, and the SHAP method is adopted to analyze the model's feature importance and reveal its decision-making mechanism, thus solving the "black box" problem of deep learning models. Experimental results on the carbon fiber reinforced composite plate guided wave dataset show that the proposed method achieves an average classification accuracy of 93.62%, with all evaluation metrics significantly outperforming the single-branch 1D-CNN, LSTM and Transformer benchmark models. SHAP analysis verifies that time domain features are the core discriminant features of the model, and the model's decision-making logic is consistent with the physical mechanism of Lamb wave propagation in defective composite materials. The research shows that time-frequency feature fusion can fully mine defect discriminant information in Lamb wave signals, the multi-branch 1D-CNN is well adapted to guided wave signal feature learning, and SHAP effectively realizes the physical interpretability of the model. This method provides an interpretable technical approach for intelligent non-destructive testing of composite materials and has good engineering application potential.

Fucang Zhao, Qingjun Feng. Lamb Wave Damage Identification Method for Composites Based on Dual-domain Features and Interpretable Deep Network. Academic Journal of Computing & Information Science (2026), Vol. 9, Issue 3: 105-111. https://doi.org/10.25236/AJCIS.2026.090313.

[1] B. F. Junqueira, R. Leiderman, D. A. Castello, Damage recovery in composite laminates through deep learning from acoustic scattering of guided waves, Ultrasonics 139 (2024) 107293.

[2] M. Tanveer, M. U. Elahi, J. Jung, M. M. Azad, S. Khalid, H. S. Kim, Recent advancements in guided ultrasonic waves for structural health monitoring of composite structures, Applied Sciences 14 (2024) 11091.

[3] C. Zhang, A. A. Mousavi, S. F. Masri, G. Gholipour, K. Yan, X. Li, Vibration feature extraction using signal processing techniques for structural health monitoring: A review, Mechanical Systems and Signal Processing 177 (2022) 109175.

[4] H. Sun, W. Shao, J. Song, X. Yang, Y. Wang, X. Qing, Lamb wavefield based monogenic signal processing for quantifying delamination in composite laminates, Smart Materials and Structures 31 (2022) 105030.

[5] P. Kashyap, K. Shivgan, S. Patil, B. R. Raja, S. Mahajan, S. Banerjee, S. Tallur, Unsupervised deep learning framework for temperature compensated damage assessment using ultrasonic guided waves on edge device, Scientific Reports 14 (2024) 3751.

[6] D. Perfetto, A. De Luca, M. Perfetto, G. Lamanna, F. Caputo, Damage detection in flat panels by guided waves based artificial neural network trained through finite element method, Materials 14 (2021) 7602.

[7] S. Zhao, P. Shi, D. Han, A novel mechanical fault signal feature extraction method based on unsaturated piecewise tri-stable stochastic resonance, Measurement 168 (2021) 108374.

[8] F. Zhang, K. Zhang, H. Cheng, D. Gao, K. Cai, Ultrasonic guided wave health monitoring of high-temperature aircraft structures based on variational mode decomposition and fuzzy entropy, in: Actuators, volume 13, MDPI, 2024, p. 411.

[9] S. Huang, H. Sun, S. Wang, K. Qu, W. Zhao, L. Peng, Sswt and vmd linked mode identification and time-of-flight extraction of denoised sh guided waves, IEEE Sensors Journal 21 (2021) 14709–14717.

[10] J. Moll, J. Kathol, C.-P. Fritzen, M. Moix-Bonet, M. Rennoch, M. Koerdt, A. S. Herrmann, M. G. Sause, M. Bach, Open guided waves: online platform for ultrasonic guided wave measurements, Structural Health Monitoring 18 (2019) 1903–1914.