本文主要應用Elman動態類神經對模鑄式比壓器的老化狀態軌跡進行預測，並以實際量測驗證之。為探討模鑄式比壓器老化特性與局部放電間的關係，本研究在控制雜訊干擾的環境下，以過電壓的方式加速比壓器老化，利用高頻比流器(HFCT)、耦合電容(TEV)及超高頻感測器(UHF)感測模鑄式比壓器(3.3-6 kV級中壓比壓器)接地線上的電流脈衝訊號，以NI PXI-5105的高速資料擷取卡(DAQ)進行資料擷取與傳輸到高頻寬嵌入式控制器NI PXIe-8135，將量測資料整理成電壓值百分之九十的幅値(統計結果顯示大多局部放電訊號出現再0.9倍左右)及保留在取樣週期中重複超過兩次的放電信號相角做為老化指標，由於絕緣老化時間非常長，為了不必要浪費的記憶空間，探討最佳感測器及取樣的時間間隔分別為一分鐘、十五分鐘及三十分鐘進行討論，最後以HFCT感測器及採取三十分鐘時間間隔量測一次進行存取。趨勢分析上使用小波分解方式進行資料前處理剔除高頻成份，保留低頻成份的平穩趨勢數據，接著將兩種特徵參數輸入Elman動態類神經網路做判讀處理。藉由分析HFCT每間隔三十分鐘取得的局部放電訊號，每十二小時評估一次模鑄式比壓器的絕緣狀態，對Elman動態類神經預測系統的性能進行驗證。

This study aims to predict the aging trend of Cast-Resin Potential Transformer with Elman Dynamic Neural Network, and to verify by field measurements, so as to further investigate the correlation between the aging characteristic of dry type voltage transformer and partial discharge phenomenon. The tests were performed in a noise/interference free environment by adjusting voltages to accelerate the deterioration of insulation material of dry type voltage transformer at 3.3-6kV capacity and to detect current pulses from the grounding with multiple sensors, such as HFCT, TEV and UHF sensor. The measured data will be collected and transmitted, a high-bandwidth embedded host controller, and be organized with 90% of voltage amplitude and the partial discharge signal repeatedly occurred with the same phase angle more than 2 times in one sampling cycle as an aging condition. Considering a long deterioration process, the sampling rates were respectively set for 1, 15 and 30 minutes as not to waste the data storage and to achieve the most effective sensor application, it is decided ultimately to take the measurements by HFCT at 30-minute intervals. First using wavelet analysis to filter out high frequency bandwidth and to obtain the low frequency with stable trending, the two characteristics will be transmitted to Elman dynamic neural network for further diagnosis and analysis of the partial discharge signals sampling every 30 minutes with HTCT as well as assessment of insulation condition for dry type voltage transformer every 12 hours. In addition, a specific algorithm was designed and installed onto a human machine interface to verify the performance of Elman dynamic neural network.

[1]蘇經洲、戴政祺、陳建富、殷偉綸、洪銘良、李耀郎，「具音波傳導功能之高壓模鑄型變壓器」，中華民國第二十七屆電力工程研討會，2006 。
[2]古峰昌，「多功能局部放電分析儀之研製及其應用」，國立台灣科技大學電機工程系博士論文，2013。
[3]李傳揚、宋建成、吝伶艷，「10 kV異步電動機定子繞組電暈放電指紋分佈規律的研究」，太原理工大學煤礦裝備與安全控制山西省重點實驗室，2014。
[4]陳耿堯，「結合音射探頭組與單晶片微控制器檢測電路之局部放電檢測系統」，國立成功大學電機工程學系碩士論文，2013。
[5]「變壓器之選用、試驗、運轉與維護」，經濟部工業局九十一年度工業技術人才培訓計畫講義。
[6]財團法人台灣大電力研究試驗中心，「高壓變壓器局部放電現場查驗方式之研究」，2002。
[7]S. Birlasekaran, “Comparison of Known PD Signals With the Developed and Commercial HFCT Sensors”, IEEE Transactions on Power Delivery, Vol. 22, No. 3, pp. 1581–1590, 2007.
[8]“High Voltage Test Techniques - Partial Discharge Measurement”, IEC 60270, 2000.
[9]J. M. Mart´ınez-Tarifa, G. Robles, M. V. Rojas-Moreno, and J. Sanz-Feito, “Partial Discharge Pulse Shape Recognition Using An Inductive Loop Sensor”, Measurement Science And Technology, Vol. 21, No. 10, pp. 1–10, 2010.
[10]A. Eigner and K. Rethmeier, “An Overview on the Current Status of Partial Discharge Measurements on AC High Voltage Cable Accessories”, IEEE Electrical Insulation Magazine, Vol. 32, No. 2, 2016.
[11]D. A. Nattrass, “Partial Discharge Measurement and Interpretation”, IEEE Electrical Insulation Magazine, Vol. 4, No. 3, pp.10–23, 1988.
[12]K. Goto and H. Shiota, “Lifetime Diagnosis of Transformer,” OHM., Vol. 1097, No. 11, pp. 4348, 2000.
[13]T. Boczar, “Identification of a Specific Type of PD from Acoustic Emission Frequency Spectra, ”IEEE Transactions on Dielectrics and Electrical Insulation,Vol. 8, No. 4, pp. 598–606, 2001.
[14]E. Gulski, “Digital Analysis of Partial Dischares, ”IEEE Transactions on Dielectrics and Electrical Insulation, Vol. 2, No. 5, pp. 822–837, 1995.
[15]H. Borsi, “The Relation between Thermal and Electrical Stress and the PD Behavior of Epoxy-Resin Transformers, ”IEEE Transactions on Electrical Insulation, Vol. 28, No. 6, pp. 1007–1015, 1993.
[16]謝毓城，「電力變壓器手冊」，保定天威保變電氣股份有限公司，2014。
[17]M. S. Naidu and V. Kamaraju, 「High Voltage Engineering」,McGraw Hill Education Offices, 1996.
[18]蕭如泉、何金良，「高電壓電工學」，中國水利水電出版社，1997。
[19]賴耿陽，「環氧樹脂應用實務」，復漢出版社，1998。
[20]周澤存，「高電壓技術」，中國電力出版社，1988。
[21]V. K. Agarwal, H. M. Banford, B. S. Bernstein, E. L. Brancato, R. A. Fouracre, G. C. Montanari, J. L. Parpal, J. L. Parpal, J .N. Seguin, D.M. Ryder, and J. Tanaka, “The Mysteries of Multifactor Ageing,” IEEE Transactions on Electrical Insulation Magazine, Vol. 11, No. 3, pp. 37–43, 1995.
[22]E. L. Brancato, “An Updata on Multifactor Ageing, “IEEE Transactions on Electrical Insulation Magazine, Vol. 14, No. 5, pp.22–23, 1998.
[23]T. Zhao and R. A. Bernstorf, “Ageing Tests of Polymeric Housing Materials for Non-Ceramic Insulators,” IEEE Transactions on Electrical Insulation Magazine, Vol. 14. No. 2. pp.26–33, 1998.
[24]G. C. Montanari, A. Motori, A. T. Bulinski, S. S. Bamji, and J. Densley, “Application of Oxidation Induction Time and Compensation Effect to the Diagnosis of HV Polymeric Cable Insulation,” IEEE Transactions on Diagnosis on Dielectrics and Electrical Insulation,Vol.3, No. 3, pp.351–360, 1996.
[25]K. Udayakumar, M. A. Panneerselvam and K. Dharmalingam, “Endurange Estimation of Epoxy Resin Systems under Synergistic Environment of High Relative Humidity and Electrical Stress,” Proceeding of the 3^rd International Conference on Properties and Applications of Dielectric Materials, pp. 711–714, Tokyo, Japan, 1991.
[26]National Instrument “NI PCI-5105, 60MS/s, 12-bit, 8-channel digitizer/oscilloscope,” 2013.
[27]葉宇利，「5.3 GHz昇頻器與降頻器設計與製作」，國立中央大學通訊工程學系碩士論文，2003。
[28]Ming Ren , Ming Dong,Zhong Ren , Hua-Dong Peng and Ai-Ci Qiu,“Transient Earth Voltage Measurement in PD Detection of Artificial Defect Models in SF6,”IEEE Transactions on Plasma Science, Vol. 40, pp. 2002–2008, 2012 .
[29]Xuefeng Zhao, Lu Pu , Shuangzan Ren, Zeli Ju and Wenyan Zhao, “Partial discharge characteristics and mechanism in SF6 Gas under oscillating impulse voltages,” Meas. Sci. Technol., Vol. 22, No.3, pp. 35704–35 710, 2011.
[30]Ming Ren, Ming Dong and Aici Qiu,“Partial discharges in S6 gas filled void under standard aperiodic and oscillating switching impulses,” IEEE Transactions on Dielectrics and Electrical Insulation,Vol. 21, pp. 262–272, 2014.
[31]A. Rodrigo, P. Llovera, V. Fuster and A. Quijano , “Influence of high frequency current transformers bandwidth on charge evaluation in partial discharge measurements”, IEEE Transactions on Dielectrics and Electrical Insulation, Vol. 18, No. 5.1798–1802, 2011.
[32]R. M Rao and A. S. Bopardikar, “Wavelet Transforms:Introduction to Teory and Applications,” Addison Wesley, first printing, 1998.
[33]胡昌華、張軍波、夏軍及張偉等，「基於MATLAB的系統分析與設計-小波分析」，西安電子科技大學出版社，2000。
[34]林美妍，「基於經驗模態分解的局部放電信號小波分析」，國立成功大學電機工程學系碩士論文，2011。
[35]X. Ma, C. Zhou, and I. J. Kemp, “Interpretation of Wavelet Analysis and Its Application in Partial Discharge Detection, ” IEEE Transactions on Dielectrics and Electrical Insulation, Vol. 9, No. 3, pp. 446–457, 2002.
[36]I. Shim, J. J. Soraghan and W. H. Siew, “Detection of PD utilizing digital signal processing methods part 3: open-loop noise reduction,” IEEE Electrical Insulation Magazine, Vol. 17, No. 1, pp. 6–13, 2001.
[37]“Instrument Transformers-part1: Current Transformers,” IEC44-1, IEC Central Office,Geneva,Switzerland, 1996.
[38]D. Paul, “Failure Analysis of Dry-Type Power Transformer, “IEEE Transactions on Industry Applications, Vol. 37, No. 3, pp. 689–695, 2001.
[39]G. Keppel and S. Zedeck, “Data Analysis for Research Designs-Analysis of Variance and Multiple Regression/Correlation Approaches,” Freeman and Company, 1989.
[40]張斐章、張麗秋，「類神經網路導論原理與應用」，滄海圖書資訊股份有限公司，2015。
[41]王進德，「類神經網路與模糊控制理論入門與應用」，全華圖書股份有限公司，2015。
[42]Yván Túpac, Álvaro Talavera and Cristian Rodríguez Rivero,“Dynamic and recursive oil-reservoir proxy using Elman neural networks,” IEEE Andescon,pp.1–4,2016.
[43]W. Z. Sun and J. S. Wang, “Elman Neural Network Soft-Sensor Model of Conversion Velocity in Polymerization Process Optimized by Chaos Whale Optimization Algorithm,” IEEE Journals & Magazines, Vol. 5, pp. 13062–13076, 2017.
[44]Chih-Min Lin and Enkh-Amgalan Boldbaatar,“Fault Accommodation Control for a Biped Robot Using a Recurrent Wavelet Elman Neural Network,” IEEE Journals & Magazines, Vol. 11, No. 4, pp. 2882–2893, 2017.
[45]鄭允中，「基於長短期記憶遞迴類神經網路之新台幣兌台幣美元匯率預模型」，國立台灣大學電機資訊學院工程學系碩士論文，2017。