本論文為應用渦電流檢測技術來建立一套非破壞性金屬厚度檢測系統，能有效運用於PCB銅箔製程，量測銅箔厚度以避免因厚度不均而造成電特性改變，並透過KNN模型分類製程PCB板時鑽孔下的銅箔及未鑽孔的銅箔類別，以梯度提升決策樹(GBDT)來回歸探頭所量測到之銅箔厚度。於K最近鄰分類法(KNN)上，根據距離較近的鄰居之類型，判別其資料的類別為何，藉由探頭所得振幅及相位值對每個經過鑽孔的銅箔及未經過鑽孔的銅箔標籤並預測其所屬類別，避免將量測到之鑽孔點視為厚度上的瑕疵。最後藉由GBDT結合眾多弱學習器來擬合成一強學習器，預測銅箔厚度值，此方法能同時增加檢測銅箔範圍提高檢測速度並保有一定準確度以穩定系統。在線圈感測器部分，本論文透過FEM模擬建立一理想模型環境並分析線圈對不同軟性銅箔基板(Flexible Copper Clad Laminate, FCCL)之變化，從中找尋一最佳諧振點及鑑別度之線圈作為本論文系統之感測器，可有效檢測出不同厚度以及不同數量的鑽孔點情形使線圈阻抗及相位產生變化，最後以實際量測來驗證模擬的可行性並分析其結果。

This paper is to establish a non-destructive metal film thickness testing system using eddy current detection technology, which can be effectively applied in the copper of PCB’s process and measure thickness of copper to avoid changing its electrical properties due to rugged thickness. This paper uses the K-Nearest Neighbor (KNN) classification model to classify the drilled and non-drilled copper, using GBDT to regress the thickness of copper measured by the probe. KNN determines the type of samples according to the Euclidean distance of k nearest neighbors. As a result, avoiding recognizing the drilled copper as defects of thickness, this paper uses the amplitude and phase obtained by the probe to label each drilled and non-drilled copper. Then the KNN predicts their belonging classification. Finally, this paper also uses the GBDT to combine many weak learners and fits them into a strong learner to regress the thickness of copper. This method can increase the detection range of copper, detection speed, and maintain a reliable accuracy to stabilize the system. In the part of sensor, this paper uses FEM simulation to model an idea environment and analyze the coil's variances of different Flexible Copper Clad Laminates (FCCL). Finding a great coil with the best resonance point and distinction as a sensor can also effectively detect different thicknesses and numbers of drilled copper with the changes of impedance and phase. Finally, this paper verifies the feasibility of the simulation by using the actual measurement.

[1]C. V. Dodd and W. E. Deeds, “Analytical Solutions to Eddy-Current Probe-Coil Problems”, Journal of Applied Physics, Vol. 39, pp. 2829-2838, 1968.
[2]J. C. Moulder, Uzal.E, and J. H. Rose, “Thickness and conductivity of metallic layers from eddy current measurements”, Review of Scientific Instruments, Vol. 63, pp. 3455-3465, Jan. 1992.
[3]C. C. Tai and J. H. Rose, “Thickness and conductivity of metallic layers from pulsed eddy-current measurements”, Review of Scientific Instruments, Vol. 61, pp. 3965-3972, Aug. 1996.
[4]C. C. Tai, H. C. Yang, and Y. H. Lin, “Modeling the surface condition of ferromagnetic metal by the swept-frequency eddy current method”, IEEE Transactions on Magnetics, Vol. 38, pp. 205-210, Jan. 2002.
[5]C. C. Tai, “Advanced eddy-current methods for quantitative NDE ” Department of Electrical Engineering Iowa State University Dissertation for Doctor of Philosophy, 1997.
[6]H. C. Yang and C. C. Tai, “Pulsed eddy-current measurement of a conducting coating on a magnetic metal plate ”, Measurement Science and Technology, Vol. 13, No. 8, pp. 1259-1265, July. 2002.
[7]C. C. Tai, “Characterization of coatings on magnetic metal using the swept frequency eddy current method”, Review of Scientific Instruments, Vol. 71, pp. 3161-3167, Aug. 2000.
[8]戴政祺、楊弘吉，「渦電流探頭尺寸與金屬夾層瑕疵檢測信號之關係探討」， 技術學刊，Vol. 17, No. 4, pp. 525-533, 2002.
[9]C. C. Tai and S. F. Wang, “Time-Domain and Frequency-Domain Eddy Current Simulations by the Finite Element Method”, Key Engineering Materials, Vol. 270-273, pp. 585-592, Jan, 2004.
[10]王勝豐，「有限元素法分析掃頻式和脈衝式渦電流檢測」，國立成功大學電機工程學系碩士論文，2003。
[11]Y. L. Pan and C. C. Tai, “Thickness and Conductivity Analysis of Molybdenum Thin Film in CIGS Solar Cells Using Resonant Electromagnetic Testing Method”, IEEE Transactions on Magnetics, Vol. 48, No.2, pp. 347-350, Feb. 2012.
[12]潘彥霖，「光感影像法分析之多重物理量數值建模與分析:缺陷與渦電流探頭磁場之檢測」，國立成功大學電機工程學系博士論文，2010。
[13]翁新發，「基於諧振原理之渦電流檢測儀設計與應用」，國立成功大學電機工程學系碩士論文，2012。
[14]李柏毅，「非接觸式微米級金屬薄膜檢測系統研製及田口法最佳感測線圈設計」 ， 國立成功大學電機工程學系碩士論文，2017。
[15]蔡侑侖，「非接觸式微米級印刷電路板銅箔厚度量測系統及多品質特性之最佳感測線圈設計」，國立成功大學電機工程學系碩士論文，2018。
[16]林牧穎，「基於支持向量回歸渦電流金屬薄膜厚度檢測系統設計與評估」，國立成功大學電機工程學系碩士論文，2019。
[17]R. Smid, A. Docekal, and M. Kreidl, “Automated classification of eddy current signatures during manual inspection”, NDT&E International, Vol. 38, No. 6, pp. 462-470, Sept. 2005.
[18]J. Park, S. J. Han, N. Munir, Y. T. Yeom, S. J. Song, H. J. Kim, and S. G. Kwon, “MRPC eddy current flaw classification in tubes using deep neural networks” , Nuclear Engineering and Technology, Vol. 51, No. 7, pp. 1784-1790, Oct. 2019.
[19]A. Bernieri, L. Ferrigno, M. Laracca, and M. Molinara, “Crack Shape Reconstruction in Eddy Current Testing Using Machine Learning Systems for Regression”, IEEE Transactions on Instrumentation and Measurement, Vol. 57, No. 9, pp. 1958-1968, Sept. 2008.
[20]A. Skarlatos and T. Theodoulidis, “Calculation of the Eddy-Current Flow Around a Cylindrical Through-Hole in a Finite-Thickness Plate”, IEEE Transactions on Magnetics, Vol. 51, No. 9, pp. 1-7, Sept. 2015.
[21]R. Hamia, C. Cordier, and C. Dolabdjian, “Separability of Multiple Deep Crack Defects With an NDE Eddy Current System”, IEEE Transactions on Magnetics, Vol. 49, No. 1, pp. 124-127, Jan. 2013.
[22]L. S. Rosado, F. M. Janeiro, P. M. Ramos, and M. Piedade, “Defect Characterization With Eddy Current Testing Using Nonlinear-Regression Feature Extraction and Artificial Neural Networks”, IEEE Transactions on Instrumentation and Measurement, Vol. 62, No. 5, pp. 1207-1214, May. 2013.
[23]J. Martinos, T. Theodoulidis, N. Poulakis, and A. Tamburrino, “A Benchmark Problem for Eddy Current Nondestructive Evaluation”, IEEE Transactions on Magnetics, Vol. 50, No. 2, pp. 1053-1056, Feb. 2014.
[24]X. Mao and Y. Lei, “Thickness measurement of metal pipe using swept-frequency eddy current testing”, NDT&E International, Vol. 78, pp. 10-19, March. 2016.
[25]A.Bülent Doyum and Serkan Ertekin, “A practical hole testing technique using a meter display eddy current instrument” NDT&E International, Vol. 31, No. 1, pp. 11-15, Feb. 1998.
[26]Jerome H. Friedman, “Greedy Function Approximation: A Gradient Boosting Machine”, The Annals of Statistics, Vol.29, No.5, pp.1189-1232, Oct. 2001.
[27]H. Ding, G. Li, X. Dong, and Y. Lin, “Prediction of Pillar Stability for Underground Mines Using the Stochastic Gradient Boosting Technique”, IEEE Access, Vol. 6, pp. 69253-69264, Nov. 2018.
[28]T. Chen et al., “Prediction of Extubation Failure for Intensive Care Unit Patients Using Light Gradient Boosting Machine”, IEEE Access, Vol. 7, pp. 150960-150968, Oct. 2019.
[29]B. Zhang, J. Ren, Y. Cheng, B. Wang, and Z. Wei, “Health Data Driven on Continuous Blood Pressure Prediction Based on Gradient Boosting Decision Tree Algorithm”, IEEE Access, Vol. 7, pp. 32423-32433, March. 2019.
[30]戴政祺，「非破壞性檢測」上課講義，國立成功大學電機系。
[31]“AD9851 Datasheet” Analog Devices, Inc.
[32]“AD637 Datasheet” Analog Devices, Inc.
[33]賴行健，「倒傳遞類神經網路及主成份分析應用於金屬扣件硬度篩選檢測:系統設計與評估」，國立成功大學電機工程學系碩士論文，2019。
[34]陳林湋，「低頻渦電流金屬厚度檢測系統研製」，國立成功大學電機工程學系碩士論文，2012。
[35]劉尹雄，「脈衝式渦電流檢測系統之設計與應用」，國立成功大學電機工程學系碩士論文，1999。
[36]Trevor Hastie, Robert Tibshirani, and Jerome Friedman, “The Elements of Statistical Learning”, Springer, 2008.
[37]K. Tsukada, M. Hayashi, Y. Nakamura, K. Sakai, and T. Kiwa, “Small Eddy Current Testing Sensor Probe Using a Tunneling Magnetoresistance Sensor to Detect Cracks in Steel Structures”, IEEE Transactions on Magnetics, Vol. 54, No. 11, pp. 1-5, Nov. 2018.
[38]D. Wen, M. Fan, B. Cao, B. Ye, and G. Tian, “Lift-Off Point of Intersection in Spectral Pulsed Eddy Current Signals for Thickness Measurement”, IEEE Sensors Letters, Vol. 2, No. 2, pp. 1-4, June. 2018.
[39]H. Wang, W. Li, and Z. Feng, “Noncontact Thickness Measurement of Metal Films Using Eddy-Current Sensors Immune to Distance Variation”, IEEE Transactions on Instrumentation and Measurement, Vol. 64, No. 9, pp. 2557-2564, Sept. 2015.
[40]T. Cover and P. Hart, “Nearest Neighbor Pattern Classification”, IEEE Transactions on Information Theory, Vol. 13, No. 1, pp. 21-27, Jan. 1967.
[41]A. E. Lakhdari, A. Cheriet, and I. N. El-Ghoul, “Skin effect based technique in eddy current non-destructive testing for thickness measurement of conductive material”, IET Science, Measurement & Technology, Vol. 13, No. 2, pp. 255-259, Feb. 2019.
[42]C. A. Stott, P. R. Underhill, V. K. Babbar, and T. W. Krause, “Pulsed Eddy Current Detection of Cracks in Multilayer Aluminum Lap Joints”, IEEE Sensors Journal, Vol. 15, No. 2, pp. 956-962, Feb. 2015.
[43]李航，「統計學習方法」，清華大學出版社，2012。
[44]L. Breiman, J.H. Friedman, R.A. Olshen, and C.J. Stone, “Classification and Regression Trees”, Taylor & Francis, 1984.
[45]Leo Breiman, “Arcing the Edge”, Technical Report 486, Statistics Department University of California at Berkeley, 1997.
[46]L. Mason, J. Baxter, P.Bartlett, and M. Frean, “Boosting Algorithms as Gradient Descent”, Neural Information Processing Systems, Vol.12, pp. 512-518, Nov. 1999.