本論文突破傳統超聲波換能器 「一打一收」的工作模式，以P(VDF-TrFE) 為壓電材料設計並製作出一種新形態的聚焦式超聲波換能器，名其為解析式超聲波換能器，此類換能器的聚焦鏡頭上經由黃光微影、金屬蒸鍍與壓電薄膜極化的步驟，佈置多個獨立的壓電感測單元，使其具備更佳的時間與空間解析能力，可同時獲得不同位置的背向散射聲場資訊，因此得以更加掌握缺陷特徵，解決傳統超聲波換能器單一壓電元件感測面積的訊號積分現象。
實驗部分，本論文使用解析式超聲波換能器配合本實驗室建立的多通道擷取系統，首先對設計之「垂直表面裂縫深度漸變試片」執行C-Scan掃描，並引用卷積神經網路模型代替人為辨識，將於此試片收集的大量背向散射訊號資訊轉為二維影像投入人工智慧模型進行訓練，使模型擁有辨識垂直表面裂紋深度的能力。同時，針對自行設計之「介面缺陷試片」，同樣進行C-Scan掃描，以獲得之波型訊號資料繪製掃描區域的灰階影像，以測試所研發之解析式超聲波換能器的感知介面缺陷能力。
本研究基於背向散射聲場是入射聲場與試片內部結構交互作用的結果，提出以解析式超聲波換能器取代傳統超聲波換能器，對表面裂縫與介面缺陷進行掃描，並以卷積神經網路模型取代人為檢測，減少人力資源耗費的同時，也提升了超聲波檢測的精確度，更可以藉由更多聲場資訊還原出缺陷的重要資訊。

This study overcame the “Through-Transmission” mode of conventional ultrasonic transducers by a self-made new type of transducer named analytic ultrasonic transducer, based on P(VDF-TrFE) piezoelectric solution. The focusing lens of this type of transducer undergoes processes such as photolithography, electron beam evaporation, and polarization to arrange multiple independent piezoelectric sensing units, enhancing both temporal and spatial resolution, allowing the simultaneous acquisition of backscattered acoustic field information from different locations. As a result, it is more capable of capturing defect characteristics and overcomes the signal integration effect caused by single sensing area of conventional ultrasonic transducers.
The analytic ultrasonic transducer is utilized together with a developed multi-channel acquisition system for C-Scan. First, implement C-Scan on a designed specimen with gradient-depth vertical surface crack. Convolutional neural network (CNN) model is then employed to replace manual recognition. The vast amount of backscattered signal data collected from the specimen is converted into 2D images and fed into the artificial intelligence model for training, enabling it to identify the depth of vertical surface cracks. Meanwhile, implement C-Scan on designed interfacial defect specimens, then generate grayscale images of the scanned area by the collected waveform data to ensure the capability of analytic ultrasonic transducer to detect interfacial defects.
This study is based on the understanding that backscattered acoustic field results from the interaction between incident acoustic field and the internal structure of the specimen. Analytical ultrasonic transducer is employed to replace conventional ones for scanning surface cracks and interfacial defects. Additionally, using convolutional neural network instead of manual inspection, which not only reduces labor costs but also enhances the accuracy of ultrasonic evaluation. By utilizing more acoustic field information, it is able to reconstruct critical details about defects.

H. Kawai, “The Piezoelectricity of Poly (vinylidene Fluoride),” Jpn. J. Appl. Phys., vol. 8, pp. 975-976, 1969.
D. Xiang, N. H. Hsu, and G. V. Blessing, “The design, construction and application of a large aperture lens-less line-focus PVDF transducer,” Ultrasonics, vol. 34, pp. 641-647, 1996.
S. Smoloz, and W. Grill, “Focusing PVDF Transducers for Acoustic Microscopy,” Res. Nondestr. Eval., vol. 7, pp. 195-201, 1996.
W. Zou, S. Holland, K.Y. Kim, and W. Sachse, “Wideband high-frequency line-focus PVDF transducer for materials characterization,” Ultrasonics, vol. 41, pp. 157-161, 2003.
M. Robert, G. Molingou, K. Snook, J. Cannata, and K. Shung, “Fabrication of focused poly (vinylidene fluoride-trifluoroethylene) P(VDF-TrFE) copolymer 40-50 MHz ultrasound transducers on curved surfaces,” J. Appl. Phys., vol. 96, no. 1, pp. 252-256, 2004.
C. H. Chung, and Y. C. Lee, “Fabrication of poly(vinylidene fluoride-trifluoroethylene) ultrasound focusing transducers and measurements of elastic constants of thin plates,” NDT&E Int., vol. 43, pp. 96-105, 2010.
A. Habib, S. Wagle, A. Decharat, and F. Melandsø, “Evaluation of adhesive-free focused high-frequency PVDF copolymer transducers fabricated on spherical cavities,” Smart Mater. Struct., vol. 29, no. 4, 045026, 2020.
A. Masnata, and M. Sunseri, “Neural network classification of flaws detected by ultrasonic means,” NDT&E Int., vol. 29, pp. 87-93, 1996.
S. W. Liu, Jin. H. Huang, J.C. Sung, and C.C. Lee, “Detection of cracks using neural networks and computational mechanics,” Comput. Methods Appl. Mech. Engrg, vol. 191, pp. 2831-2845, 2002.
N. Munir, H. J. Kim, J. Park, S. J. Sung, and S. S. Kang, “Convolutional neural network for ultrasonic weldment flaw classification in noisy conditions,” Ultrasonics, vol. 94, pp. 74-81, 2019.
I. Virkkunen, T. Koskinen, O. Jessen-Juhler, and J. Rinta-Aho, “Augmented ultrasonic data for machine learning,” J. Nondestruct. Eval., vol. 40, pp. 1-11, 2021.
O. Siljama, T. Koskinen, O. Jessen-Juhler, and I. Virkkunen, “Automated Flaw Detection in Multi-channel Phased Array Ultrasonic Data Using Machine Learning,” J. Nondestruct. Eval., vol. 40, pp. 1-13, 2021.
H.-H Huang, and Y.-C. Lee, “Fabrication and Characterization of PVDF Hydrophone,” M.S. Thesis, Department of Mechanical Engineering, National Cheng Kung University, Taiwan, 2006.
J.-L Chiu, and Y.-C. Lee, “PVDF hydrophones and Analytical Ultrasound Transducer,” M.S. Thesis, Department of Mechanical Engineering, National Cheng Kung University, Taiwan, 2007.
C.-R Jiang, and Y.-C. Lee, “The Development of Linear PZT Hydrophone Array and Rapid Ultrasound Scanning System,” M.S. Thesis, Department of Mechanical Engineering, National Cheng Kung University, Taiwan, 2010.
C.-T Kuo, and Y.-C. Lee, “An FPGA-Based Ultrasonic Image Scanning System and Non-immersed Ultrasonic Inspection,” M.S. Thesis, Department of Mechanical Engineering, National Cheng Kung University, Taiwan, 2014.
Y.-Y. Wei, and Y.-C. Lee, “Line-Focused P(VDF-TrFE) Transducer for Measuring Lanb Wave Propagation on Plates with Periodic Structures,” M.S. Thesis, Department of Mechanical Engineering, National Cheng Kung University, Taiwan, 2022.
T.-P. Huang, and Y.-C. Lee, “Lamb Wave and Piezoelectric Transducers Applied to Ultrasonic Touch Panel with Artificial Intelligence,” M.S. Thesis, Department of Mechanical Engineering, National Cheng Kung University, Taiwan, 2020.
C.-Y. Chiang, and Y.-C. Lee, “Differential Sensing of Lamb Wave Using Piezoelectric Transducer for Baseline Free Tactile Sensing on a Plate,” M.S. Thesis, Department of Mechanical Engineering, National Cheng Kung University, Taiwan, 2023.
C.-S. Lo, and Y.-C. Lee, “Fabrication and Application of Analytic Back Scattering Arrayed Transducer,” M.S. Thesis, Department of Mechanical Engineering, National Cheng Kung University, Taiwan, 2013.
M.-C. Hsieh, and Y.-C. Lee, “Analytic Ultrasound Transducer and Artificial Intelligence for Ultrasonic Non-Destructive Evaluation,” M.S. Thesis, Department of Mechanical Engineering, National Cheng Kung University, Taiwan, 2023.
M. Wang, S. Lu, D. Zhu, J. Lin, and Z. Wang, “A High-Speed and Low-Complexity Architecture for Softmax Function in Deep Learning,” IEEE Asia-Pacific Conference on Circuits and Systems, 2018.
Pieter-Tjerk de Boer, D. P. Kroese, S. Mannor, and R. Y. Rubinstein, “A Tutorial on the Cross-Entropy Method,” Annals of Operations Research, vol. 134, no.1, pp. 19-67, 2005.
D. P. Kingma, J. Ba, “Adam: A Method for Stochastic Optimization,” 3^rd International Conference on Learning Representations, San Diego, California, May. 7-9, 2015.
J. A. Bonnell, “Implementation of a New Sigmoid Function in Backpropagation Neural Networks,” Electronic Theses and Dissertations, pp.1342, 2011.