定制化小波基赋能的薄热障涂层太赫兹可解释测厚新方法

doi:10.3901/JME.2025.24.012

摘要/Abstract

摘要： 薄热障涂层的陶瓷层厚度远小于太赫兹(Terahertz, THz)系统中心波长，导致THz信号严重混叠，难以准确提取飞行时间与折射率，使测厚误差增大。为此，提出小波基赋能的薄热障涂层THz可解释测厚新方法，创新设计定制化高斯小波解混叠层，自主构建与反射峰特征相似的小波基，以消除信号混叠，使可解释网络结构能够准确提取测厚所需的飞行时间与折射率。首先，建立考虑热障涂层结构的THz信号解析模型，探索反射峰的形状特征，以此为模板，优选适用于解混叠的小波族。其次，提出定制化高斯小波解混叠层，自主构建出与前两峰高度相似的基函数，准确分离反射峰，并建立稀疏层筛选峰值关键信息，提升解析模型生成的仿真训练集与试验测试集一致性。然后，构建物理可解释测量模块分别从时、频域特征中提取飞行时间与折射率，将两者结果以除法方式连接，求解准确陶瓷层厚度。最后，制备热障涂层样品，开展THz试验，结果表明：构建的定制化高斯小波基与反射峰相关系数超过0.93，所提出方法与已有的六种现有测厚算法相比，其精度最高，最大测厚误差小于5 μm，耗时为8.2 ms。

关键词: 热障涂层, 太赫兹无损检测, 定制小波基, 可解释神经网络, 解析模型

Abstract: The thickness of topcoat in thin thermal barrier coating is much smaller than the centre wavelength of the terahertz (THz) system, generating seriously overlapped THz signals. The time of flight and refractive index are difficult to be extracted accurately, which increases the error of thickness measurement. To this end, a novel method is proposed for THz interpretable thickness measurement of thin thermal barrier coatings empowered by wavelet bases, and a customized wavelet de-aliasing layer is innovatively designed to autonomously construct wavelet bases with similar characteristics of the peaks, followed by separating overlapped THz signals and enabling the interpretable network structure to accurately extract time-of-flight and refractive index for thickness measurement. Firstly, an analytical model of THz signals considering the structure of thermal barrier coating is established to explore the shape characteristic of peak as a template for selecting preferred wavelet family for signal de-aliasing. Secondly, a customized Gaussian wavelet de-aliasing layer is developed to autonomously construct a wavelet basis function which is almost identical with the first two peaks to accurately separate the peaks of THz signals. A sparse layer is established to select the key information of peaks and improve the agreement between the simulated training set from analytical model and the experimental test set. Then, a physically interpretable measurement module is constructed to extract time-of-flight and refractive index from the time and frequency domain features, and these two results are connected by a division layer to solve topcoat thickness. Finally, thermal barrier coatings were prepared and THz experiments were carried out. The results show that the correlation coefficient between the proposed customized Gaussian wavelet basis and THz peaks exceeds 0.93, and the proposed method is superior to six existing thickness measurement methods, with the maximum thickness error of less than 5 μm and the time consumed being 8.2 ms.

Key words: thermal barrier coating, terahertz nondestructive testing, customized wavelet basis, analytical model, interpretable neural network

中图分类号:

V263

孙凤山, 范孟豹, 曹丙花, 叶波. 定制化小波基赋能的薄热障涂层太赫兹可解释测厚新方法[J]. 机械工程学报, 2025, 61(24): 12-27.

SUN Fengshan, FAN Mengbao, CAO Binghua, YE Bo. Customized Wavelet Basis Enabled Novel Terahertz Interpretable Thickness Measurement method of Thin Thermal Barrier Coatings[J]. Journal of Mechanical Engineering, 2025, 61(24): 12-27.

参考文献

[1] 张平,钟舜聪,张秋坤,等. 热障涂层折射率与厚度的太赫兹无损检测[J]. 机械工程学报,2021,57(20):47-56. ZHANG Ping,ZHONG Shuncong,ZHANG Qikukun,et al. Terahertz nondestructive testing of refractive index and thickness of thermal barrier coating[J]. Journal of Mechanical Engineering,2021,57(20):47-56.
[2] PADTURE N P. Advanced structural ceramics in aerospace propulsion[J]. Nature Materials,2016,15(8):804-809.
[3] GUO W,ZHAO,Y,QIAO L,et al. High precision thickness evaluation of thermal barrier coating with high frequency eddy current testing method[J]. NDT & E International,2023,140(11):1-7.
[4] QIAN C,KOU X,PEI C X,et al. Topcoat thickness measurement of thermal barrier coating using grating laser acoustic spectrum method[J]. Ceramics International,2022,48(3):3676-3684.
[5] LI C,FAN M B,CAO B H,et al. Thickness measurement of thermal barrier coating based on mutual inductance of eddy current system[J]. IEEE Transactions on Industrial Electronics,2024,71(7):8099-8109.
[6] 林莉,马志远,孙珞茗,等. 航空功能涂层表界面完整性参数超声检测反演原理及其应用[J]. 机械工程学报,2023,59(8):1-8. LIN Li,MA Zhiyuan,SUN Luoming,et al. Inversion principle and application of ultrasonic detecting surface or interface integrity parameters of aviation functional coatings[J]. Journal of Mechanical Engineering,2023,59(8):1-8.
[7] HUANG Z,ZHU J G,ZHUO L J,et al. Non-destructive evaluation of uneven coating thickness based on active long pulse thermography[J]. NDT & E International,2022,130(5):1-8.
[8] LUO L R,CHEN Y,ZHOU M,et al. Progress update on extending the durability of air plasma sprayed thermal barrier coatings[J]. Ceramics International,2022,48(13):18021-18034.
[9] 刘增华,吴育衡,王可心,等. 基于太赫兹时域光谱技术的陶瓷基复合材料缺陷检测成像研究[J]. 机械工程学报,2023,59(14):33-42. LIU Zenghua,WU Yuheng,WANG Kexin,et al. Research on defect detection imaging of ceramic matrix composites based on terahertz time-domain spectroscopy technology[J]. Journal of Mechanical Engineering,2023,59(14):33-42.
[10] LIEBERMEISTER L,NELLEN S,KOHLHAAS R B,et al. Optoelectronic frequency-modulated continuous- wave terahertz spectroscopy with 4THz bandwidth[J]. Nature Communications,2022,12(1):1-10.
[11] CUI Y Q,XU Y F,HAN D H,et al. Hidden-information extraction from layered structures through terahertz imaging down to ultralow SNR[J]. Science Advances,2023,9(40):8435-8446.
[12] LUO M T,ZHONG S C,YAO L G,et al. Thin thermally grown oxide thickness detection in thermal barrier coatings based on SWT -BP neural network algorithm and terahertz technology[J]. Applied Optics,2020,59(13):4097-4104.
[13] UNNIKRISHNAKURUP S,DASH J,RAY S,et al. Nondestructive evaluation of thermal barrier coating thickness degradation using pulsed IR thermography and THz-TDS measurements:A comparative study[J]. NDT & E International,2020,116(11):1-9.
[14] SUN F S,FAN M B,CAO B H,et al. High-resolution terahertz imaging of debonding defects in thermal barrier coatings using an optimal wavelet transform[J]. IEEE Transactions on Instrumentation and Measurement,2023,72(8):1-11
[15] GE H,SUN Z,LU X,et al. THz spectrum processing method based on optimal wavelet selection[J]. Optics Express,2014,3(32):4457-4472.
[16] XU Y F,WANG X Y,ZHANG L Y,et al. Terahertz nondestructive quantitative characterization for layer thickness based on sparse representation method[J]. NDT & E International,2021,124(9):1-7.
[17] MEI H W,JIANG H Y,YIN F H,et al. Detection of small defects in composite insulators using terahertz technique and deconvolution method[J]. IEEE Transactions on Instrumentation and Measurement,2020,69(10):8146-8155.
[18] DAI B,WANG P,WANG T Y,et al. Improved terahertz nondestructive detection of debonds locating in layered structures based on wavelet transform[J]. Composite Structures,2017,168(3):562-568.
[19] NSENGIYUMVA W,ZHONG S C,LUO M T,et al. Terahertz spectroscopic characterization and thickness evaluation of internal delamination defects in GFRP composites[J]. Chinese Journal of Mechanical Engineering,2023,36(1):1-21.
[20] ZHANG Z H,PENG G Z,TAN Y P,et al. THz wave detection of gap defects based on convolutional neural network improved by residual shrinkage network[J]. CSEE Journal of Power and Energy Systems,2023,9(3):1078-1089.
[21] HONG H P,CUI X Z,QIAO D,et al. Simulating nonstationary non-Gaussian vector process based on continuous wavelet transform[J]. Mechanical Systems and Signal Processing,2022,165(2):1-19.
[22] WANG M H,LU S D,LIAO R M,et al. Fault diagnosis for power cables based on convolutional neural network with chaotic system and discrete wavelet transform[J]. IEEE Transactions on Power Delivery,2021,37(1):582-590.
[23] WADDIE A J,SCHEMMEL P J,CHALK C,et al. Terahertz optical thickness and birefringence measurement for thermal barrier coating defect location[J]. Optics Express,2020,28(21):31535-31552.
[24] CAO B H,DENG T,FAN M B,et al. Physical constraints-based terahertz thickness measurement method of thermal barrier coating[J]. NDT & E International,2024,143(4):1-12.
[25] KRIMI S,TOROSYAN G,BEIGANG R. Advanced GPU-based terahertz approach for in-line multilayer thickness measurements[J]. IEEE Journal of Selected Topics in Quantum Electronics,2017,23(4):1-12.
[26] ZHANG H Z,SHI L L,HE M X. Extension of terahertz time-domain spectroscopy:A micron-level thickness gauging technology[J]. Optics Communications,2022,506(6):1-7.
[27] FUKUCHI T,FUSE N,OKADA M,et al. Measurement of refractive index and thickness of topcoat of thermal barrier coating by reflection measurement of terahertz waves[J]. Electronics and Communications in Japan,2013,96(12):702-708.
[28] 徐亚飞,王星雨,钟舜聪,等. 基于稀疏表示的叶片热障涂层太赫兹检测回波信号定位分离方法[J]. 机械工程学报,2023,59(8):9-19. XU Yafei,WANG Xingyu,ZHONG Shuncong,et al. Location and separation of terahertz echo signal of TBCs based on sparse representation method[J]. Journal of Mechanical Engineering,2023,59(8):9-19.
[29] SUN F S,FAN M B,CAO B H,et al. Terahertz based thickness measurement of thermal barrier coatings using long short-term memory networks and local extrema[J]. IEEE Transactions on Industrial Informatics,2022,18(4):2508-2517.
[30] LI X R,WILLIAMS B M,MAY R K,et al. Optimizing terahertz waveform selection of a pharmaceutical film coating process using recurrent network[J]. IEEE Transactions on Terahertz Science and Technology,2022,12(4):392-400.
[31] SANCHEZ A,HESHMAT B,AGHASI A,et al. Terahertz time-gated spectral imaging for content extraction through layered structures[J]. Nature Communications,2016,7(9):1-7.
[32] GONG Y L,CAO B H,ZHANG H,et al. Terahertz based thickness measurement of thermal barrier coatings using hybrid machine learning[J]. Nondestructive Testing and Evaluation,2023:1-17(2023-12-4),https://doi.org/10.1080/10589759.2023.2167991
[33] SUN F S,CAO B H,FAN M B,et al. Physics-constrained transfer learning for self-assessment of terahertz thickness measurement of thermal barrier coatings[J]. IEEE Transactions on Industrial Informatics,2024,20(9):11045-11056.
[34] YE D D,WANG W Z,YIN C D,et al. Pulsed terahertz spectroscopy combined with hybrid machine learning approaches for structural health monitoring of multilayer thermal barrier coatings[J]. Optics Express,2020,28(21):34875-34893.
[35] TU W L,ZHONG S C,SHEN Y C,et al. Neural network-based hybrid signal processing approach for resolving thin marine protective coating by terahertz pulsed imaging[J]. Ocean Engineering,2019,173(3):58-67.
[36] 孙凤山,范孟豹,曹丙花,等. 基于时频关键信息融合的热障涂层太赫兹准确测厚方法[J]. 机械工程学报,2023,59(14):10-22. SUN Fengshan,FAN Mengbao,CAO Binghua,et al. Terahertz based accurate thickness measurement of thermal barrier coatings using the key time-frequency information fusion[J]. Journal of Mechanical Engineering,2023,59(14):10-22.
[37] XU Y F,ZHOU H K,CUI Y Q,et al. Full scale promoted convolution neural network for intelligent terahertz 3D characterization of GFRP delamination[J]. Composites Part B:Engineering,2022,242(8):1-13.
[38] ZHANG Z H,DING D W,WANG L M,et al. THz wave defect detection technology based on attention autoencoder and semisupervised ladder network[J]. IEEE Sensors Journal,2023,23(8):8961-8972.
[39] RUDIN C. Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead[J]. Nature Machine Intelligence,2019,1(5):206-215.
[40] SUN H Y,PENG L S,LIN J J,et al. Development of a physics-informed doubly fed cross-residual deep neural network for high-precision magnetic flux leakage defect size estimation[J]. IEEE Transactions on Industrial Informatics,2022,18(3):1629-1640.
[41] SUN F S,FAN M B,CAO B H,et al. THzResNet:A physics-inspired two-stream residual network for thermal barrier coating thickness measurement[J]. IEEE Transactions on Industrial Informatics,2023,19(7):8328-8339.
[42] 严如强,周峥,杨远贵,等. 可解释人工智能在工业智能诊断中的挑战和机遇:归因解释[J].机械工程学报,2024,60(12):21-40. YAN Ruqiang,ZHOU Zheng,YANG Yuangui,et al. Challenges and opportunities of xai in industrial intelligent diagnosis:Attribution interpretation[J]. Journal of Mechanical Engineering,2024,60(12):21-40.
[43] SHANG Z G,ZHAO Z B,YAN R Q,et al. Denoising fault-aware wavelet network:A signal processing informed neural network for fault diagnosis[J]. Chinese Journal of Mechanical Engineering,2023,36(1):1-18.
[44] LI T F,ZHAO Z B,SUN C,et al. WaveletKernelNet:An interpretable deep neural network for industrial intelligent diagnosis[J]. IEEE Transactions on Systems Man Cybernetics-Systems,2022,52(4):2302-2312.
[45] LI T F,SUN C,LI S A,et al. Explainable graph wavelet denoising network for intelligent fault diagnosis[J]. IEEE Transactions on Neural Networks and Learning Systems,2024,35(6):8535-8548.
[46] LI S A,LI T F,SUN C,et al. WPConvNet:An interpretable wavelet packet kernel-constrained convolutional network for noise-robust fault diagnosis[J]. IEEE Transactions on Neural Networks and Learning Systems,2024,35(10):14974-14988.
[47] NOURINOVIN S,RAHMAN M M,PARK S J,et al. Terahertz dielectric characterization of three-dimensional organotypic treated basal cell carcinoma and corresponding double Debye model[J]. IEEE Transactions on Terahertz Science and Technology,2023,13(3):246-253.
[48] SUN F S,CAO B H,FAN M B,et al. THzMINet:A terahertz model-data-driven interpretable neural network for thickness measurement of thermal barrier coatings[J]. IEEE Transactions on Industrial Informatics,2024,20(3):4722-4734.