本論文採用創新的方法所建立之多層支撐向量機網路模型（Multi-Layer Support Vector Machine，SVM）做為電力變壓器之故障診斷。利用解碼技巧將株落選擇演算法應用於改善向量機之精確度，這在文獻上是首次呈現。利用特徵向量及徑向基函數（Radial Basis Function，RBF）參數選擇做為診斷出發生故障的先期徵兆，以改善向量機之精確度。本研究所提出方式是可以把干擾向量機之不必要特徵向量去除，並且將RBF參數做最佳化。
藉由本論文所建立的多層支撐向量機網路模型（Multi-Layer Support Vector Machine，SVM）結合株落選擇演算法（Clonal Selection Algorithm ，CSA）診斷出發生故障的先期徵兆，以便進一步停機進行相關維護程序，以有效降低變壓器維護成本，並期延長變壓器運轉壽命與提升變電所之供電可靠度。
最後本論文係以高雄港務局第五貨櫃中心變電站之變壓器進行實測與資料收集，並將蒐集所得油中氣體成分資料，佐以IEC 60599資料庫及台電歷年來之故障案例樣本資料，以所提之多層支撐向量機網路模型進行實際案例診斷，驗證系統之可用性。經由實際資料分析結果，本研究所提出方式是更有效率及可行性，並且操作更快速及增加向量機之精確度。

This thesis presents an innovative method based on multi-layer Support Vector Machine (SVM) combined with Clonal Selection Algorithm （CSA） for the purpose of fault diagnosis of power transformers. A clonal selection algorithm (CSA) based encoding technique is applied to improve the accuracy of classification, which demonstrated in the literature for the first time. With features and RBF kernel parameters selection to predict incipient fault of power transformer improve the accuracy of classification systems and the generalization performance. The proposed approach is distinguished by removing redundant input features that may be confusing the classifier and optimizing the selection of kernel parameters.
As a result, the proposed approach can assist the maintenance of power transformers and extend their operation life, as well as enhance their reliability. In order to effectively and reliably monitor power transformers in a substation, the Support Vector Machine is employed to develop Multi-Layer SVM Classifier based on pattern recognition and fault diagnosis system in the proposed approach.
Finally, the collected data associated with both cases in IEC 60599 and historical data in both Taipower system and the fifth container center of Kaohsiung port are selected for computer simulation to demonstrate the effectiveness of the proposed multi-layer SVM classifier. Simulation results of practice data demonstrate the effectiveness and high efficiency of the proposed approach, which makes operation faster and also increases the accuracy of the classification.

中文摘要 --------------------------------------------- Ⅰ
英文摘要 --------------------------------------------- Ⅱ
誌謝 --------------------------------------------- Ⅳ
目錄 --------------------------------------------- Ⅴ
圖目錄 --------------------------------------------- Ⅷ
表目錄 --------------------------------------------- Ⅹ
第一章 緒論------------------------------------- --- 1
1.1 前言----------------------------------------- 1
1.2 研究目的------------------------------------- 3
1.3 變壓器故障類型與特徵氣體及其比值間關係------- 3
1.4 國內外有關變壓器故障診斷之研究情況----------- 7
1.4.1 振動監測技術--------------------------------- 7
1.4.2 油中氣體分析技術----------------------------- 9
1.4.3 油中氣體分析之變壓器故障診斷----------------- 10
1.5 論文架構------------------------------------- 13
第二章 株落選舉演算法介紹--------------------------- 14
2.1 前言----------------------------------------- 14
2.2 免疫反應------------------------------------- 14
1.2.1 抗原----------------------------------------- 14
2.2.2 抗體----------------------------------------- 14
2.2.3 B淋巴細胞----------------------------------- 16
2.2.4 T淋巴細胞----------------------------------- 17
2.3 株落選擇與記憶細胞--------------------------- 20
2.4 株落選擇演算法------------------------------- 21
2.4.1 免疫演算法之組成架構與演算步驟--------------- 21
2.4.2 株落選擇演算法之組成架構--------------------- 22
2.4.3 株落選擇演算法之演算步驟--------------------- 24
第三章 支撐向量機理論------------------------------- 27
3.1 支撐向量機的基本原理------------------------- 28
3.2 支撐向量機的回歸算法及其實現----------------- 33
3.2.1 線性迴歸問題--------------------------------- 33
3.2.2 非線性迴歸問題------------------------------- 36
3.3 核函數--------------------------------------- 39
3.4 徑向基底函數--------------------------------- 41
3.4.1 徑向基底網路的特性--------------------------- 43
3.4.2 徑向基底網路的架構--------------------------- 43

3.4.3 高斯函數------------------------------------- 47
第四章 建構變壓器故障診斷系統----------------------- 49
4.1 建構支撐向量機分類器------------------------- 49
4.1.1 輸入向量------------------------------------- 49
4.1.2 支撐向量------------------------------------- 50
4.1.3 核函數的選取--------------------------------- 50
4.2 回歸模型的參數選取----------------------- --- 51
4.3 以多層支撐向量機分類器為基礎的變壓器故障診斷- 54
4.3.1 支撐向量機的輸入向量------------------------- 54
4.3.2 支撐向量機網路的訓練------------------------- 54
4.3.3 支撐向量機網路的測試驗證--------------------- 59
4.4 多層支撐向量機分類器網路主程式執行流程------- 60
4.5 建構最佳訓練參數之求解策略------------------- 62
第五章 診斷實例與結果分析--------------------------- 66
5.1 IEC TC 10 資料庫故障案例分析----------------- 66
5.2 SVM網路模型與BPNN網路性能比較--------------- 79
5.3 SVMs與ANNs配合CSA的結果分析---------------- 82
5.3.1 當採用IEC TC 10 資料庫----------------------- 82
5.3.2 當採用台電資料庫----------------------------- 82
5.4 診斷實例的結果分析--------------------------- 83
第六章 結論----------------------------------------- 94
6.1 具體成果------------------------------------- 94
6.2 未來的研究方向------------------------------- 96
參考文獻 --------------------------------------------- 98
附錄一 IEC TC 10 資料庫(Faulty Equipment)----------- 102
附錄二 IEC TC 10 資料庫(Typical(Normal) Values)----- 107
附錄三 TPC&KHB故障案例統計表----------------------- 110

[1]IEC Publication 599, “Interpretation of The Analysis of Gases in Transformers And Other Oil-Filled Electrical Equipment In Service,” Fisrt Edition, 1978.
[2]“IEEE Guide for The Interpretation of Gases Generated In Oil-Immersed Transformers,” IEEE Std.C57.104-1991.
[3]R.R. Rogers, “IEEE And IEC Codes to Interpret Faults in Transformers Using Gas in Oil Analysis,” IEEE Trans Electron. Insulat. 13 (5), pp.349–354,1978.
[4]M.H. Wang, “A Novel Extension Method for Transformer Fault Diagnosis,” IEEE Trans. Power Deliv. 18 (1),pp.164–169,2003.
[5]N.C.Wang,”Development of Monitoring and On-line Diagnosing System for Large Power Transformers,”（Power Reserch Institute,TPC）.
[6]Yuanping.Ni, “One Intelligent Method for Fault Diagnosis and Its Application,” IEEE Trans. Power Deliv, pp755-758,2000.
[7]C.L.Wang,”支撐向量機在地下水位預報中的應用研究,”（2002-05-29）（http://www.cws.net.cn/Journal/slxb/200305/22.html）
[8]”BECTA”,Power Station Magazine in Russia,No.6,1998。
[9]C. E. Lin, J. M. Ling, and C. L. Huang, “An Expert System for Transformer Fault Diagnosis and Maintenance Using Dissolved Gas Analysis,”IEEE Trans. Power Delivery, vol. 8, pp. 231–238, Jan. 1993.
[10]Y. C. Huang, H. T. Yang, and C. L. Huang, “Developing A New Transformer Fault Diagnosis System Through Evolutionary Fuzzy Logic,” IEEE Trans. Power Delivery, vol. 12, pp. 761–767, Apr. 1997.
[11]H. T. Yang and C. C. Liao, “Adaptive Fuzzy Diagnosis System for Dissolved Gas Analysis of Power Transformers,” IEEE Trans. Power Delivery,vol. 14, pp. 1342–1350, Oct. 1999.
[12]Y. Zhang, X. Ding, Y. Liu, and P. J. Griffin, “An Artificial Neural Network Approach to Transformer Fault Diagnosis,” IEEE Trans. Power Delivery, vol. 11, pp. 1836–1841, Oct. 1997.
[13]Z.Wang, Y. Liu, and P. J. Griffin, “A Combined ANN and Expert System Tool for Transformer Fault Diagnosis,” IEEE Trans. Power Delivery, vol.13, pp. 1224–1229, Oct. 1998.
[14]C.H. Yann, “Evolving Neural Nets for Fault Diagnosis of Power Transformers,” IEEE Trans. Power Del. 18 (3), pp.843–848,2003.
[15]M.H. Wang, “Extension Neural Network for Power Transformer Incipient Fault Diagnosis,” IEE Proceedings-Generation, Trans. Distrib. 150 (6) ,pp.679–685,2003.
[16]Yann-Chang Huang, Chao-Ming Huang, “Evolving Wavelet Networks for Power Transformer Condition Monitoring,” IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 17, NO. 2, pp.412-416, APRIL 2002.
[17]Z.Y. Wang, Y.L. Liu,and P.J. Griffin, “A Combined ANN and Expert System Tool for Transformer Fault Diagnosis,” IEEE Trans. Power Del.13 (4) pp.1224–1229, 1998.
[18]K.F. Thang, R.K. Aggarwal, A.J. McGrail, and D.G. Esp, “Analysis of Power Transformer Dissolved Gas Data Using The Self-organizing Map,” IEEE Trans. Power Del. 18 (4) pp.1241–1248, 2003.
[19]J. J. Kelly, “Transformer Fault Diagnosis By Dissolved-gas Analysis,” IEEE Trans. Ind. Applicat., vol. 16, pp. 777–782, Nov. 1980.
[20]E. Dornenburg and W. Strittmater, “Monitoring Oil Cooling Transformers by Gas Analysis,” Brown Boveri Rev., vol. 61, pp. 238–274, 1974.
[21]“Interpretation of The Analysis of Gases in Transformers and Other Oil filled Electrical Equipment in Service,” IEC Pub. 599, 1978.
[22]Ganyun Lv., Haozhong Cheng, Haibao Zhai, and Lixin Dong, “Electric Power Systems Research 75,” 9–15, 2005 Elsevier B.V. All rights reserved.
[23]C. Cortes, and V. Vapnik, “Support Vector Networks,” Mach Learning 20 (3) pp.273–295, 1995.
[24]J.W. Lu, K.N. Plataniotis, and A.N. Venetsanopoulos, “Face Recognition Using Feature Optimization and Support Vector Learning Neural Networks for Signal Processing XI,”Proceedings of the 2001 IEEE Signal, Processing Society Workshop, 10–12 pp. 373–382, September 2001.
[25]F.E.H. Tay, and L.J. Cao, “Application of Support Vector Machines in Financial Time Series Forecasting,” Omega, August, 29 (4), pp.309–317, 2001.
[26]W.W. Yan, and H.H. Shao, “Application of Support Vector Machine Nonlinear Classifier to Fault Diagnoses,” Proceedings of the Fourth World Congress Intelligent Control and Automation, 10–14 Shanghai, China, pp. 2697–2670, June 2002.
[27]L.B. Jack, and A.K. Nandi, “Fault Detection Using Support Vector Machines and Artificial Neural Networks: Augmented by Genetic Algorithms,”Mech. Syst. Signal Process. 16 (2–3), pp.373–390, 2002.
[28]W.C. Chan, C.W. Chan, K.C. Cheung, and C.J. Harris, “On The Modeling of Nonlinear Dynamic Systems Using Support Vector Neural Networks,” Eng. Appl. Artif. Intell. 14, pp.105–113,2001.
[29]V.N. Vapnik, “The Nature of Statistical Learning Theory,” Springer-Verlag, New York, 1995.
[30]Vapnik V N. The Nature of Statistical Learning Theory. NY: Springer-Verlag, 1995.
[31]張學工, ”關於統計學習理論與支撐向量機”, [J].自動化學報 pp.32-42, 2000.
[32]Vapnik V, Golowich S,and Smola A. “Support Vector Method for Function Approximation, Regression Estimation, And Signal Processing [A].” In: Mozer M, Jordan M, Petsche T (eds). Neural Information Processing Systems [M]. MIT Press, 1997, 9.
[33]李凱,郭子雪,”一種基於SVM的函數模擬方法,” [J].微機發展,pp.5-6, 2001.
[34]馬云潜,張學工,”支撐向量機函數擬合在分形插值中的應用,” [J].中國清華大學學報(自然科學版),pp.76-78,2000.
[35]Müller K-R, Smola A J,and Ra tsch G, et al. “Predicting Time Series With Support Vector Machines [Z].” In: Proc. of ICANN'97, Springer Lecture Notes in Computer Science, pp.999-1005, 1997.
[36]J.R.Huang,”Using Support Vector Machine to Category Images Variation for the Image Query,”20,July 2004(Thesis for the Degree of Master).
[37]Kuan-yu,Chen, “Application of Support Vector Regression in Forecasting International Tourism Demand,” Tourism Management Research Vol.4 No.1, pp.81~97,Jine 2004.
[38]“使用放射性基底函數進行精密運動之幾何誤差補償,”（機密加工原理http://www.me.ntou.edu.tw/~pml/err.doc）。
[39]K. K. Tan, S. N. Huang, and H. L. Seet, “Geometrical Error Compens-ation of Precision Motion System Using Radial Basis Function,” IEEE Transactions on Instrumentation and Measurement, Vol.49, No.5, pp. 984–991, 2000.
[40]趙育漢,”Geometric Analysis of Photolithography Overlay Error Using Neural Networks,” July 2003 (Thesis for the Degree of Master).
[41]M.L.Jiang, “Process Recipe Optimization in Multiple Response System,” July 2002 (Thesis for the Degree of Master).
[42]Mark J. and L. Orr, “Introduction to Radial Basis Function Networks,” April 1996.
[43]Chang,C.C.,and Lin, C.J., “LIBSVM: A Library for Support Vector Machines,”Technical Report, Department of Computer Science and Information Engineering, Nation Taiwan University, Available at http://www.csie.edu.tw/~cjlin/papers/libsvm.pdf,2001.
[44]Duan, K., Keerthi, S., and Poo, A., “Evaluation of Simple Performance Measures for Tuning SVM Hyperparameters.” Technical Report, Department of Mechanical Engineering, National University of Singapore, 2001.
[45]Michel Duval, ”Interpretation of Gas-In-Oil Analysis Using New IEC Publication 60599 and IEC TC 10 Databases,” IEEE Electrical Insulation Magazine, March/April 2001 — Vol. 17, No. 2
[46]“IEC Publication 60599,” March 1999.
[47]M,Dong, ”Fault Diagnosis Model Power Transformer Based on Statistical Learning Theory and Dissolved Gas Analysis,”IEEE 2004 Internation Symposium on Electrical Insulation,p85-88.
[48]羅華強, ”類神經網路─MATLAB的應用,”（清蔚科技出版）。
[49]Hofmeyr,S.A.,and Forrest, S., “Immunity by Design: An Aritifical Immune System”,Dept. of Computer Science,1999.
[50]Shao,X., Chen,Z.,and Lin,X., “Resolution of multicomponent overlapping chromatogram using an immune alogorithm and genetic alogorithm”,Chemometric and Intelligent Laboratory System,Vol.50,pp91-99,200.
[51]Shao,X., Yu,Z., and Sun,L., “Immune alogorithms in analytical chemistry”,Trends in Analytical chemistry, Vol.22, No.2 ,pp59-69, 2003.
[52]羅意茹, ”應用免疫演算法於可控退化率存貨問題之研究,”（嘉義大學碩士論文）。
[53]廖子銘, ”類免疫演算法於多目標最佳化問題之研究與應用,”（ 大同大學碩士論文）。