本研究是要建構一個辨識率高的模鑄式比流器局部放電圖譜辨識系統，應用類神經網路並結合二維小波轉換，能辨識出模鑄式比流器局部放電圖譜。主要研究分為局部放電圖譜實驗測試及圖譜辨識系統兩大部份，其中局部放電圖譜實驗測試部分，製造特別設計七種常發生局部放電模型的模鑄式比流器，且在屏蔽實驗室內，使用局部放電檢測儀器，獲取3D放電圖譜資料；圖譜辨識的部分，為了建立類神經網路辨識系統，首先探討實驗數據化簡的處理方式，作為類神經網路的訓練及辨識用數據，並分別探討類神經網路不同的演算法、隱藏層的神經元數目，經研究後實驗數據化簡為60x60矩陣時，採用最大值模式比採用平均值模式總辨識率較高，而類神經網路採用比例共軛梯度演算法比有彈性倒傳遞演算法總辨識率較高，隱藏層的神經元數目採用20時，可建立一個辨識率98.6 %的辨識系統。為了再提高類神經網路辨識系統的辨識率，結合二維小波轉換及不同的小波函數進行特徵擷取，作為倒傳遞類神經網路之訓練集，用以建構辨識率更佳的模鑄式比流器局部放電圖譜辨識系統。研究後採用最佳60x60矩陣經二維小波轉換後，小波函數採用Coif1時，以第二層的低頻影像小波係數，隱藏層的神經元數目30至60時，再經過類神經網路辨識系統可達到100%的辨識率。最後探討加入模擬雜訊後的辨識結果，模擬雜訊在30pC以下時模鑄式比流器良品的辨識率達到100%，因此研究後之模鑄式比流器局部放電圖譜的辨識系統可用於模鑄式比流器製造廠的品質管制使用，確保出廠的模鑄式比流器皆為良品。
關鍵字：局部放電、比流器、類神經網路、小波轉換

The study pertains to developing a high recognition resin cast current transformers’ partial discharge pattern recognition system, adopting an artificial neural network and also combining the two-dimensional wavelet transform, which can identify resin cast current transformers’ partial discharge patterns. The study is mainly divided into the two parts of partial discharge pattern experiment testing and pattern recognition system; of which, on partial discharge pattern experiment testing, resin cast current transformers specially designed with seven commonly occurred partial charge models are fabricated and tested using partial discharge testing and screening instruments at insulated laboratories to derive the 3D partial discharge pattern data. On pattern recognition, for instilling an artificial neural recognition system, the study first explores the simplified processing method for the experiment data for artificial neural network training and data recognition, and also explores separately the artificial neural network’s various algorithms, neural node count on the hidden layers. After the study simplifies the experiment data into a 60x60 matrix, it is found that the maximum value model has a higher recognition ratio than the median value model, and that the artificial neural network combining the conjugate gradient algorithm has a higher recognition ratio than the resilient back-propagation algorithm; when the neural node count is set to 20 for the hidden layers, it is able to instill a recognition system with a 98.6% successful recognition ratio. To further improve the artificial neural network recognition system’s recognition ratio, two-dimensional wavelet transform is combined with varied resonant functions to derive the characteristics as the reverse transgression neural network’s training set, with which to instill an optimal recognition ratio of the resin cast current transformer’s partial discharge pattern recognition system. The study then adopts an optimized 60x60 matrix, which is put through two-dimensional wavelet transform, where the wavelet function is set to Coif1, using the second tier’s low-frequency image resonant wavelet index, and when the hidden layers neural node count reaches between 30 and 60, the artificial neural network recognition system is able to achieve a 100% recognition ratio. Lastly, by exploring the recognition results after adding simulated noise signals, it is found that the resin cast current transformer is able to achieve a 100% recognition ratio when the simulated noise signals are kept to under 30pC, and thus validating that the study’s resin cast current transformer’s partial discharge pattern recognition system can be used in quality control at resin cast current transformer manufacturing factories to ensure that all resin cast current transformers shipped from the factory are of good quality products.
Keywords: Partial Discharge, Current Transformer, Artificial Neural Network, Wavelet Transform

[1] 成永紅，「電力設備絕緣檢測與診斷」，中國電力出版社，第1-5頁 (2000)。
[2] 吳瑞南、林育勳，「大型電力變壓器預防診斷量測技術介紹」，量測資訊，第83期，第12-14頁 (2002)。
[3] 楊正光、林育勳、吳瑞南和張宏展，「高壓變壓器局部放電現場查驗方式之研究」，經濟部能源研究發展基金計畫之期末報告，台北 (2002)。
[4] Borsi, H., “The Relation between Thermal and Electrical Stress and the PD Behavior of Epoxy-Resin Transformers” , Journal of the IEEE Transactions on Electrical Insulation, Vol. 28, No. 6, pp. 1007-1015 (1993)
[5] Agarwal, V. K., “The Mysteries of Multifactor Ageing” , Journal of the IEEE Electrical Insulation Magazine, Vol. 11, No. 3, pp. 37-43 (1995)
[6] 張宏展、吳瑞南、陳在相和蕭弘清，「受電設備型式認證及檢驗機構暨原製造廠商認可標準及程序之研究」，經濟部能源研究發展基金計畫之期末報告，台北 (2001)。
[7] “High-Voltage Test Techniques-Partial Discharge Measurements” , IEC 60270, pp. 1-12 (2001)
[8] Gulski, E., “Digital Analysis of Partial Discharges” , Journal of the IEEE Transactions on Dielectrics and Electrical Insulation, Vol. 2, No. 5, pp. 822-837 (1995)
[9] Kreuger, F. H., Gulski, E., and Krivda, A., “Classificaction of Partial Discharge” , Journal of the IEEE Transactions on Electrical Insulation, Vol. 28, No. 6, pp. 917-931 (1993)
[10] Krivda, A., “Automated Recognition of Partial Discharge” , Journal of the IEEE Transactions on Dielectrics and Electrical Insulation, Vol. 2, No. 5, pp. 796-821 (1995)
[11] Kexiong, K. G., Li, T. F., and Wu, C., “PD Pattern Recognition for Stator Bar Models with Six Kinds of Characteristic Vectors Using BP Network” , Journal of the IEEE Transactions on Electrical Insulation, Vol. 9, No. 3, pp. 381-389 (2002)
[12] Wang, Z., Li, C., Peng, P., Ding, L., Jia, Y., Wang, W., and Wang, J. C., “Partial Discharge Recognition of Stator Winding Insulation Based on Artificial Neural Network,” IEEE Conference on Electrical Insulation, pp. 9-12 (2000)
[13] 許晉維，「應用類神經網路於比流氣局部放電圖譜之辨識」，碩士論文，國立台灣科技大學，台北 (2003)。
[14] 林漢偉，「應用二維小波轉換與類神經網路於模鑄式比流器局部放電圖譜之辨識」，碩士論文，國立台灣科技大學，台北 (2004)。
[15] Kuo, C. C., “Artificial Identification System for Transformer Insulation Aging” , Journal of the Expert System with Applications 37, pp. 4190-4197 (2010)
[16] 廖財昌，高壓電絕緣工程，全華科技圖書股份有限公司，台北 (1998)。
[17] 顏世雄，高電壓工程，新學識文教出版中心，台北 (1990)。
[18] 邱乾政，「子絕緣特性診斷之研究」，碩士論文，國立台北科技大學，台北 (2002)。
[19] Krivda, A., “Automated Recognition of Partial Discharge” , Journal of the IEEE Transactions on Dielectrics and Electrical Insulation, Vol. 2, No. 5, pp. 796-821 (1995)
[20] Nattrass, D. A., “Partial Discharge Measurement and Interpretaion” , Journal of the IEEE Electrical Insulation Magazine, Vol. 4, NO. 3, pp. 10-23 (1988)
[21] 李建明、朱康，高壓電氣設備試驗方法，中國電力出版社，第二版，第298-322頁 (2000)。
[22] 林育勳，「數位局部放電量測應用於模鑄式比流器絕緣狀態之評估」，博士論文，國立台灣科技大學，台北 (2004)。
[23] Kreuge, F. H., Partial Discharge Detection in High Voltage Equipment, Butterworths (1989)
[24] 楊正光、林育勳，「局部放電技術應用於電力設備之預防診斷」，電機月刊，第134期，第200-206頁 (2002)。
[25] 陳昭榮、邱乾政、葉崇源和林育勳，「部份放電法應用於電力設備絕緣檢測」，中華民國第二十二屆電力工程研討會，高雄，第811-815頁 (2001)。
[26] 吳瑞南、林育勳，「局部放電電氣信號量測技術介紹」，量測資訊，第85期，第35-38頁 (2002)。
[27] 「局部放電測量及超聲波故障定位技術資料彙編(一)」，電力供應部武漢高壓研究所 (1997)。
[28] Su, C. C., Tai, C. C., Chen, C. Y., Hsieh, J. C., and Chen, J. F., “Partial Discharge Detection using Acoustic Emission Method for a Waveguide Functional High-Voltage Cast-Resin Dry-Type Transformer,” 2008 International Conference on Condition Monitoring and Diagnosis, pp. 21-24 (2008)
[29] 「局部放電量測」，中國大陸國家標準GB 7354-87。
[30] 「電力設備局部放電現場測量導則」，中國大陸電力行業標準DL 417-91。
[31] 楊正光、林育勳、吳瑞南和張宏展，「高壓變壓器局部放電現場查驗方式之研究」，九十一年度經濟部能源研究發展基金計畫之期末報告，台北 (2002)。
[32] 簡年伯，「高電壓模鑄式比流器絕緣劣化特徵辨識之研究」，碩士論文，國立台灣科技大學，台北 (2003)。
[33] 潘彥竹，「局部放電量測系統之研製」，碩士論文，國立台灣科技大學，台北 (2004)。
[34] “Operating Instructions” , Partial Discharge Detector TE 571 (1997)
[35] “Instrument Transformers Part1: Current Transformers”, IEC 60044-1, pp. 29-31 (1996)
[36] 「變比器」，中華民國國家標準CNS 11437，C4435，第51-52頁 (2001)。
[37] 葉怡成，類神經網路模式應用與實作，儒林圖書股份有限公司，第五版，台北 (1997)。
[38] 葉怡成，應用類神經網路，儒林圖書股份有限公司，第三版，台北 (2002)。
[39] 蔡英男，「應用影像處理與類神經網路於偏光膜瑕疵辨識」，碩士論文，國立台灣科技大學，台北 (2003)。
[40] 焦成李，神經網路系統理論，儒林圖書股份有限公司，台北 (2001)。
[41] Zurada, J. M., Introduction to Artificial Neural System, West Publishing Co. St. Paul, MN, USA (1992)
[42] 林昇甫、洪成安，神經網路入門與圖樣辨識，全華圖書股份有限公司，修訂二版，台北，第1-20頁 (2002)。
[43] Hagan, M. T., Demuth, H. B., and Beale, M., Neural Network Design, PWS. Pub. Co., Har/Dis, edition, USA (1996)
[44] Riedmiller, M., and Braun, H., “A Direct Adaptive Method for Faster Backpropagation Learning: The RPROP Algorithm” , Journal of the IEEE Conference on Electrical Insulation, Vol.1, pp. 586-591 (1993)
[45] 羅華強，類神經網路－Matlab的應用，清蔚科技股份有限公司，新竹 (2001)。
[46] 尤秀慧，「應用類神經網路預測呼吸器脫離成功率」，碩士論文，慈濟大學，花蓮市 (2006)。
[47] 連國珍，數位影像技術處理，儒林圖書股份有限公司，台北 (2001)。
[48] Pun, C. M., and Lee, M. C., “Rotation-invariant Texture Classification using a Two-stage Wavelet Packet Feature Approach” , Journal of the IEE Proceedings on Vision, Image and Signal Processing, Vol. 184, pp. 422-428 (2001)
[49] Mallat, S., “Wavelets for a Vision” , Journal of the Proceeding of IEEE on Image Analysis and Interpretation, Vol. 84, No. 4, pp. 604-614 (1996)
[50] Yim, C., Klarquist, W. N., and Bovik, A. C., “Multiresolution Feature Fxtraction Based on Multifrequency Decomposition,” IEEE International Conference on Systems, pp. 2384-2389 (1994)
[51] Yuming, T., Wang, Z. D., and Crossley, P. A., “Partial Discharge Pattern Recognition Based on 2-D Wavelet Transform and Neural Network Techniques,’’ IEEE Power Engineering Society Summer Meeting, pp. 411-416 (2002)
[52] 洪穎怡，「結合小波轉換與類神經網路辨別電力開關之切換」，碩士論文，私立中原大學，桃園 (2003)。
[53] Newland, D. E., An Introduction to Random Vibrations Spectral and Wavelet Analysis Longman Scientific ＆ Technical, 3rd. Ed, England (1994)
[54] Patrick, W., and Andrew, L., “Wavelet Descriptors for Multiresolution Recognition of Handprinted Characters” , Journal of the Pattern Recognition, Vol. 28, pp. 1237-1249 (1995)
[55] 廖紹綱，數位影像處理活動Matlab，全華圖書股份有限公司，台北 (1999)。
[56] Gabor, D., “Theory of Communication” , Journal of the Institution of Electrical Engineers, Vol. 93, No. 26, pp. 429-441 (2004)
[57] Heydt, G. T., Field, P. S., Liu, C.C., Pierce, D., Tu, L., and Hensley, G., “Applications of the Windowed FET to Electric Power Quality Assessment” , Journal of the IEEE Transactions. on Power Delivery, Vol.14, No.4, pp. 1411-1416 (1999)
[58] Grossman, A., and Morlet, J., “Decomposition of Hardy Functions into Square Integrable Wavelet of Constant Shape” , Journal of the SIAM of Math. Anal., pp. 723-736 (1984)
[59] Raghuveer, R. M., and Bopardikar, A. S., Wavelat Transforms: Introduction to Theory and Application, Addison Wesley Longman, Inc (1998)
[60] Huang, C. H., Lin, C. H., and Kuo, C. L., “Choas Synchronization-based Detector for Power-quality Disturbances Classification in a Power System” , Journal of the IEEE Transactions on Power Delivery, Vol.26, No.2, pp. 944-953 (2011)