電力設備中之絕緣材料會隨著時間、環境、人為等等因素而產生老化現象，為了避免因絕緣老化所造成電力設備之突發故障，如何即時監測絕緣劣化之狀態，便成為重要之課題。
　　本文採用的分析資料，為不同試驗條件的交鏈聚乙烯試驗數據，其經過濾波、化簡和特徵萃取後，得到能代表局部放電演進的特徵參數。以此作為類神經網路之輸入資料，並將特徵曲線以分期邊界區分為劣化初期、中期與末期，做為類神經網路之目標輸出，並以倒傳遞網路以及批次訓練之方式進行網路訓練，最後使用類神經網路辨識絕緣材料之劣化狀態。為了尋找出類神經網路在不同輸入層神經元數與不同隱藏層神經元數搭配下之最佳辨識成功率，本文設定輸入層神經元數為20個至70個、隱藏層神經元數為10個至90個，利用相互搭配之方式進行網路訓練。欲尋找出最佳辨識成功率之組合。
　　研究最後以放電量總和之區域平均放電量作為局部放電之特徵參數。在絕緣劣化狀態辨識方面，本文提出以輸入層神經元個數為40筆、隱藏層神經元個數為10個之組合建議，初期辨識成功率因資料之初期成份極少，故先予以忽略；中期辨識成功率約可達到89%；末期成功率約可達到78%。在絕緣劣化分期中，中期為通知設備異常之重要時機，故應格外重視。

　　The insulation materials in electrical equipment often experience insulation degradation over time, which are caused by the environment or human. In order to avoid the sudden failure of electrical equipment by insulation breakdown, the real-time monitoring on the status of insulation degradation is always a vital research subject.
　　In this paper, the analyzed data is obtained from deterioration test of XLPE with different conditions. After being processed by filtering, reduction and feature extraction, this data becomes an effective characteristic parameter as the input data of neural network, which represents the evolution of partial discharges. The characteristic curve is divided into three stages of degradation, early, medium and final, as the output data of neural network. The training method of neural network are back propagation network(BP) and batch learning. Finally, use the neural network to identification the deteriorated state of insulating materials. In order to find out the best recognition rate of success of neural network, the numbers of input layer neurons and hidden layer neurons change from 20 to 70 and from 10 to 90, respectively.
　　This thesis utilizes the trend of change of district average discharge amount to represent the trend of the partial discharge. This thesis proposes that the number of neurons in input layer and hidden layer are 40 and 10, for training the neural network to identify the insulation degradation. The early stage data is too few to be omitted. Apart from this, the medium-term success rate of recognition is 89%, and the final-term success rate of recognition is 78%.The state of insulation degradation, medium stage is the important time to notice the failure of apparatus.

[1] 張瑞村，「數位局部放電量測應用於高壓電纜終端接頭絕緣狀態之評估」，碩士論文，國立台灣科技大學，民國九十四年。
[2] 陳振銘，「地下電纜絕緣樹枝現象之劣化趨勢研究」，碩士論文，國立台灣科技大學，民國九十八年。
[3] 張建國，「地下電纜接頭局部放電線上監控系統之研製」，碩士論文，國立台灣科技大學，民國九十五年。
[4] 簡年柏，「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，民國九十三年。
[5] “High-Voltage Test Techniques–Partial Discharge Measurements”, IEC 60270, 2000.
[6] F. H. Kreuger, “Partial Discharge Detection in High-Voltage Equipment”, Buttrtworth & Co. (Publishers) Ltd, 1989.
[7] J. Densley, “Ageing mechanisms and diagnostics for power cables - an overview”, IEEE Electrical Insulation Magazine, Vol. 17, No. 1, pp. 14-22, Feb. 2001.
[8] 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，民國九十三年。
[9] X. Ma, C. Zhou and I. J. Kemp, “Interpretation of Wavelet Analysis and Its Application in Partial Discharge Dection”, IEEE Transactions on Dielectrics and Electrical Insulation, Vol. 9, No. 3, pp. 446-457, June 2002.
[10] A. Krivda, “Automated Recognition of Partial Discharges”, IEEE Transactions on Dielectrics and Electrical Insulation, Vol. 2, No. 5, pp. 796-821, Oct. 1995.
[11] M. Mansor, A. B. A. Ghani and P. S. Ghosh, “Partial discharge pattern analysis using statistical technique in XLPE cable under various soil conditions”, 2002 Annual Report Conference on Electrical Insulation and Dielectric Phenomena, pp. 707-711, Oct. 2002.
[12] 馬秀蘭、吳德邦，統計學以Microsoft Excel為例，新文京開發出版股份有限公司，民國九十四年。
[13] 柴松林，統計學，三民書局，民國八十年。
[14] 蔡瑞煌，類神經網路概論，三民書局，民國八十四年。
[15] 焦李成，神經網路系統理論，儒林圖書有限公司，民國八十年。
[16] 羅華強，類神經網路-MATLAB的應用，高立圖書有限公司，民國九十四年。
[17] J. M. Zurada, ” Introduction to Artificial Neural Systems”, PWS Pub. Co. (Publisher), 1992.
[18] M. T. Hagan, H. B. Demuth and M. Beale, “Neural Network Design”, PWS Pub. Co. (Publisher), 1996.
[19] N. C. Sahoo and M. M. A. Salama, “Trends in partial discharge pattern classification: a survey”, IEEE Transactions on Dielectrics and Electrical Insulation, Vol. 12, No. 2, pp. 248-264, Apr. 2005.
[20] J. Li, C. Sun, S. Grzybowski and C. D. Taylor, ”Partial discharge image recognition using a new group of features”, IEEE Transactions on Dielectrics and Electrical Insulation, Vol. 13, No. 6, pp. 1245-1253, Dec. 2006.
[21] M. Hoof, B. Freisleben and R. Patsch, “PD source identification with novel discharge parameters using counterpropagation neural networks”, IEEE Transactions on Dielectrics and Electrical Insulation, Vol. 4, No. 1, pp. 17-32, Feb. 1997.
[22] A. A. Mazroua, M. M. A. Salama and R. Bartnikas, “PD pattern recognition with neural networks using the multilayer perceptron technique”, IEEE Transactions on Dielectrics and Electrical Insulation, Vol. 28, No. 6, pp. 1082-1089, Dec. 1993.
[23] 林光彥，「倒傳遞類神經網路模擬手機接收信號」，碩士論文，國立台灣科技大學，民國九十六年。