Squeezing Feature-driven Bayesian Neural Network for Damage Evaluation and Prediction of Composite Plate

doi:10.3901/JME.2025.12.001

Abstract

Abstract: Aiming at the problem of predicting the fatigue damage extension of composite laminates, a damage assessment and prediction method is proposed based on the Bayesian neural network model driven by Lamb wave squeezing features. Initially, the damage state space of the composite plate is constructed from the multi-domain features, including time-frequency domain features of Lamb wave signals, dynamic time warping features, and transfer entropy features. In order to quantitatively characterize the damage degree of the composite plate, a unified compression feature(UCF) is constructed by using the Squeezing-excitation network. Then, on the basis of the strain energy release rate model of composite laminate, the damage expansion rate and the change of strain energy release rate(CSERR) are found to be linearly related. After that, the CSERR-BNN prediction model is created by embedding the CSERR into the activation function of Bayesian neural network(BNN). This model is then utilized to evaluate and predict the evolution trends of multiple damages in the laminate. The results of finite element simulation and test data analysis of laminates made of T700G unidirectional carbon fiber prepreg showed that the CSERR-BNN prediction model, driven by the UCFs of multi-channel Lamb wave signals, can accurately predict the damage state progression trend in composite laminates with accuracy, thus tracking of the damage state and assessing their remaining useful life.

Key words: ultrasonic guided wave, composite laminates, energy release rate model, Bayesian estimation, life prediction

CLC Number:

TH39
V219

LIU Xiaofeng, ZHOU Zengliang, Zhang Chunbing, Bo Lin. Squeezing Feature-driven Bayesian Neural Network for Damage Evaluation and Prediction of Composite Plate[J]. Journal of Mechanical Engineering, 2025, 61(12): 1-11.

References

[1] 赵文滔，杨颜志，王长焕，等. 基于贝叶斯推断的复合材料层间内聚力模型参数反演[J]. 机械工程学报，2022，58(6)：110-118. ZHAO Wentao，YANG Yanzhi，Wang Changhuan，et al. Parameter inversion of interlaminar cohesion model for composite materials based on Bayesian inference[J]. Journal of Mechanical Engineering，2022，58(6)：110-118.
[2] 王丹宏，叶波，段启明，等. 基于Lamb波能量和飞行时间的碳纤维复合材料疲劳损伤成像[J]. 电子测量与仪器学报，2022，36(4)：205-213. WANG Danhong，YE Bo，DUAN Qiming，et al. Fatigue damage imaging of carbon fiber composite materials based on Lamb wave energy and flight time[J]. Journal of Electronic Measurement and Instrumentation，2022，36(4)：205-213.
[3] 许国琛，陈振华，李承庚，等. CFRP层板低能冲击损伤的非线性兰姆波特征成像[J]. 机械工程学报，2023，59(10)：40-47. XU Guochen，CHEN Zhenhua，LI Chenggeng，et al. Nonlinear Lamb wave imaging for low energy impact damage of CFRP laminates[J]. Journal of Mechanical Engineering，2023，59(10)：40-47.
[4] 刘小峰，张春兵，柏林，等. 基于递归量化观测的复合材料损伤贝叶斯估计方法[J]. 仪器仪表学报，2023，44(1)：74-84. LIU Xiaofeng，ZHANG Chunbing，BO Lin，et al. Bayesian estimation of composite structure damage based on recursive quantitative observation[J]. Chinese Journal of Scientific Instrument，2023，44(1)：74-84.
[5] Feng B，Ribeiro A L，Pasadas D J，et al. Locating low velocity impacts on a composite plate using multi-frequency image fusion and artificial neural network[J]. Journal of Nondestructive Evaluation，2022，41：34.
[6] Song R，Sun L，Gao Y，et al. Global-local feature cross-fusion network for ultrasonic guided wave-based damage localization in composite structures[J]. Sensors and Actuators，A. Physical，2023，362：114659.
[7] Liao，Yun L，Wang Y，et al. Multiscale 1-DCNN for damage localization and quantification using guided waves with novel data fusion technique and new self-attention module[J]. IEEE Transactions on Industrial Informatics，2024，20(1)：492-502.
[8] Gao F，Hua J. Damage characterization using CNN and SAE of broadband Lamb waves[J]. Ultrasonics，2022，119：106592.
[9] Zhang S，Li C M，Ye W. Damage localization in plate-like structures using time-varying feature and one-dimensional convolutional neural network[J]. Mechanical Systems and Signal Processing，2021，147：107107.
[10] Wu J，Xu X，Liu C，et al. Lamb wave-based damage detection of composite structures using deep convolutional neural network and continuous wavelet transform[J]. Composite Structures，2021，276(3)：114590.
[11] Lee H，Lim H J，Skinner T，et al. Automated fatigue damage detection and classification technique for composite structures using Lamb waves and deep autoencoder[J]. Mechanical Systems and Signal Processing，2022，163(19)：108148.
[12] GANG Y. A Bayesian approach for damage localization in plate-like structures using Lamb waves[J]. Smart Materials & Structures，2013，22(3)：35012.
[13] 刘小峰，张天瑀，韦代平，等. 复合材料基体裂纹损伤演化的自适应粒子滤波预测[J]. 机械工程学报，2024，60(18)：32-42. LIU Xiaofeng，ZHANG Tianyu，WEI Daiping，et al. Adaptive particle filtering prediction of crack damage evolution in composite materials[J]. Journal of Mechanical Engineering，2024，60(18)：32-42.
[14] Corbetta M，Sbarufatti C，Giglio M，et al. A Bayesian framework for fatigue life prediction of composite laminates under co-existing matrix cracks and delamination[J]. Composite Structures，2018，187：58-70.
[15] Eleftheroglou N，Zarouchas D，Benedictus R，et al. An adaptive probabilistic data-driven methodology for prognosis of the fatigue life of composite structures[J]. Composite Structures，2020，245：112386.
[16] Jeong Y S，Jeong M K，Omitaomu O A. Weighted dynamic time warping for time series classification[J]. Pattern Recognition，2011，44(9)：2231-2240.
[17] Schreiber，Thomas. Measuring information transfer[J]. Physical Review Letters，2000，85(2)： 461-464.
[18] Corbetta M，Sbarufatti C，Giglio M，et al. A Bayesian framework for fatigue life prediction of composite laminates under co-existing matrix cracks and delamination[J]. Composite Structures，2018，187(3)： 58-70.
[19] Zhang J，Soutis C，Fan J. Effects of matrix cracking and hygrothermal stresses on the strain energy release rate for edge delamination in composite laminates[J]. Composites，1994，25(1)：27-35.
[20] Joffe R，Varna J. Analytical modeling of stiffness reduction in symmetric and balanced laminates due to cracks in 90° layers[J]. Composites Science and Technology，1999，59(11)：1641-1652.