本研究提出一種新型非破壞性檢測技術，藉由衝擊回波法，利用敲擊機構在不同厚度之高碳鋼試片表面產生敲擊，透過金屬外罩保護式壓電感測器，接收來自試片厚度方向之回波訊號；最後應用類神經網路將回波訊號訓練成資料庫，用於辨識未知金屬板材厚度。本研究可改善過去傳統非破壞性檢測技術無法在特殊場地使用之限制，諸如：(1) 檢測時間過於冗長、(2) 檢測面積大且厚度不均、(3) 無法攜帶大量耦合劑、(4) 環境限制及測量困難等問題。
本研究成功證明具有辨識未知金屬板材厚度能力，可應用於檢測大型油槽壁厚的內側腐蝕而造成鋼板厚度變薄的情況；所設計之檢測裝置具有整體尺寸小、方便攜帶、無需攜帶耦合劑、屬於乾接觸式非破壞性檢測等特點，具有很高的應用價值。綜合上述成果，本研究為材料檢測領域提供一個全新思路，提高檢測效率與準確度。

This study proposes a novel non-destructive testing technique that utilizes the impact-echo method. By using an impact mechanism to generate impacts on the surface of high-carbon steel specimens with different thicknesses, the echo signals in the thickness direction are captured using a metal-shielded piezoelectric sensor. Finally, a neural network is employed to train the echo signals and create a database for identifying unknown metal plate thicknesses. This study improves the limitations of traditional non-destructive testing techniques, such as：(1) lengthy inspection time, (2) large and uneven inspection areas, (3) the need to carry a significant amount of coupling agent, and (4) environmental constraints and measurement difficulties, which prevent their use in special locations.
This study successfully demonstrates the ability to identify unknown metal plate thicknesses using impact-echo and metal-shielded piezoelectric sensor. It is applied to inspect the internal corrosion of large oil tank which results in decreasing of the wall thickness. It has the advantages such as small overall size, easy portability, no need for coupling agents, and dry contact non-destructive testing. These characteristics make it highly valuable for practical applications and provides a new approach for the field inspection of structural integrity with both high efficiency and accuracy.

[1] Innerspec,“NDT news: Dry-coupled ultrasonic testing,”Innerspec.com., 30-December-2015. [Online]. Available: https://www.innerspec.com/news/dry-coupled-ultrasonic-testing-nde-method.
[2] A. M. Robinson, B. W. Drinkwater, and J. Allin,“Dry-coupled low-frequency ultrasonic wheel probes: application to adhesive bond inspection,”NDT E Int., vol. 36, no. 1, pp. 27–36, 2003.
[3] C. Juan, S. Lu, Z. Hou, and Q. Xin,“A roller ultrasonic probe to detect gas-pipe’s thickness,”in 2005 IEEE International Conference on Industrial Technology, pp. 1098–1102, 2005.
[4] R. Anvo, T. P. Sattar, T.-H. Gan, and I. Pinson,“NON-DESTRUCTIVE TESTING ROBOTS (NDTBOTS) FOR IN-SERVICE STORAGE TANK INSPECTION,”in Human-Centric Robotics, pp. 301-308, 2017.
[5] Y. Lecun, L. Bottou, Y. Bengio, and P. Haffner,“Gradient-based learning applied to document recognition,”Proc. IEEE Inst. Electr. Electron. Eng., vol. 86, no. 11, pp. 2278–2324, 1998.
[6] D. E. Rumelhart and J. L. McClelland,“The programmable blackboard model of reading,”in Parallel Distributed Processing, London, England: The MIT Press, pp. 318–362, 1987.
[7] P. Y. Simard, D. Steinkraus, and J. C. Platt,“Best practices for convolutional neural networks applied to visual document analysis,”in Seventh International Conference on Document Analysis and Recognition, Edinburgh, UK, Aug. 6, 2003.
[8] N. Jmour, S. Zayen, and A. Abdelkrim,“Convolutional neural networks for image classification,”in 2018 International Conference on Advanced Systems and Electric Technologies (IC_ASET), pp. 397–402, 2018.
[9] 黃祥紋,“PVDF水下聽音器之製作與特性量測,”國立成功大學機械工程學系, 2005。
[10] B. A. Auld,“ Acoustic Fields and Wave in Solids,”John Wiley & Sons, New York, vol. 1, pp. 101-103, 1973.
[11] 邱巨霖,“壓電式PVDF水聽器與解析式超音波換能器之研究,”國立成功大學機械工程學系, 2006。
[12] J. D. Achenbach, “Wave propagation in elastic solids,” New York American Elsevier Pub. Co., 1984.
[13] 洸碩科技有限公司,“應力波檢測法,敲擊回音法非破壞檢測,應力波傳,”Cae-so.com, 2009. [Online]. Available: http://cae-so.com/NDT-1.html.
[14] 張承燊,“以壓電感測器與人工智慧為基礎之超音波觸控感測器,”國立成功大學機械工程學系, 2019。
[15] W. Mcculloch and W. Pitts,“A logical calculus of the ideas immanent in nervous activity,”Bull. Math. Biol., vol. 5, no. 4, pp. 115–133, 1943.
[16] C.-S. Chang, Y.-C. Lee, and Y.-C. Wang,“An Ultrasonic Touchscreen Sensors Based on Piezoelectric Transducers and Artificial Intelligence,”M. S. Thesis, Department of Mechanical Engineering, National Chung Kung University, Tainan, Taiwan, 2019.
[17] M. Z. Rehman, N. Mohd. Nawi, M. Arshad, and A. Khan,“Recognition of cursive Pashto optical digits and characters with trio deep learning neural network models,”Electronics (Basel), vol. 10, no. 20, p. 2508, 2021.
[18] Y.-T. Chen, Y.-C. Lee, and Y.-C. Wang,“Machine Learning Applied to Ultrasonic Touch,”M.S. Thesis, Department of Mechanical Engineering, National Chung Kung University, Tainan, Taiwan, 2019.
[19] V. Ewald, R. M. Groves, and R. Benedictus,“DeepSHM: a deep learning approach for structural health monitoring based on guided Lamb wave technique,”SPIE Smart Structure + Nondestructive Evaluation, Denver, Colorado, USA, Mar. 3-7, 2019.
[20] Wikipedia contributors,“MATLAB,”Wikipedia, The Free Encyclopedia, 2023. [Online]. Available: https://zh.wikipedia.org/w/index.php?title=MATLAB&oldid=77723769.
[21] 黃騰平,“蘭姆波與壓電換能器應用於具人工智慧之超音波觸控感測面板,”國立成功大學機械工程學系, 2020。
[22] P.-T. de Boer, D. P. Kroese, S. Mannor, and R. Y. Rubinstein,“A tutorial on the cross-entropy method,”Ann. Oper. Res., vol. 134, no. 1, pp. 19–67, 2005.
[23] D.P. Kingma, J. Ba,“Adam: A method for Stochastic Optimization,”3rd International Conference on Learning Representations, San Diego, California, Mar. 7-9, 2015.
[24] A. Krizhevsky, I. Sutskever, and G. E. Hinton,“ImageNet classification with deep convolutional neural networks,” Communications of the ACM, vol. 60, no. 6, pp. 84-97, 2017.
[25] 王柏鈞,“DNN(深度神經網路)的全面認識,”機器學習歷程, 14-Feb-2020. [Online]. Available: https://medium.com/%E6%A9%9F%E5%99%A8%E5%AD%B8%E7%BF%92%E7%9F%A5%E8%AD%98%E6%AD%B7%E7%A8%8B/dnn-%E6%B7%B1%E5%BA%A6%E7%A5%9E%E7%B6%93%E7%B6%B2%E8%B7%AF%E7%9A%84%E5%85%A8%E9%9D%A2%E8%AA%8D%E8%AD%98-ad50aa531205.