在工業界，感應馬達是大部分機械來源，難免會因長年累月的運轉或偶發的事件造成感應馬達的電氣或機械故障，感應馬達的故障不但會造成生產線的停頓，更嚴重還會造成人員安全上的威脅。感應馬達的定期保養是為了降低因嚴重故障而停機的方法；然而保養的支出佔每年設備主要投資之九成，於是若能建構一套實用的監測系統，能正確有效地辨識感應馬達故障類型，將有助於預測性的保養，並大幅提升感應馬達的保養效率與可靠度。
過去一般對感應馬達的故障辨識只考慮定載的故障狀況對頻譜振幅的影響，然而負載的變化對振幅也產生至深的影響，使得單純考慮振幅變化的故障辨識在實際應用上受到相當大的限制。不同故障狀況及負載變化都會影響頻譜的結構，對於在不同負載下的故障辨識，需要同時考慮到頻帶漂移及振幅改變等因素。本論文為了解決頻帶漂移及振幅改變的問題，分別使用頻率軸等比校正、負載分段及特徵萃取的方法處理。再分別以倒傳遞類神經網路(Back Propagation Neural Network, BPNN)及廣義迴歸類神經網路(General Regression Neural Network, GRNN)進行感應馬達故障狀況訓練及辨識。

Induction motor is the most popular machine in the industry. It is used extensively in mechanical plants, and it is un- avoidable to have the motor’s electrical and mechanical faults due to continuously operating throughout the year. Faults of motors do not only cause the production line to shut down but also imperil the personnel security. A suitable motor maintenance schedule will be a needed to decrease the machine down time. However, major investment might take up to 90% for equipment, and it would be helpful to have a practicable low-cost supervisory scheme on maintenance. If the faults of machine can be detected correctly and effectively, the maintenance efficiency and dependability could be increased greatly.
In the past, researches on fault recognition for Induction motors only concentrated on Spectrum analysis with amplitudes based on a constant load. However, the frequency and amplitude of the spectrum analyzed under different fault conditions are also affected significantly by load variations. Using spectrum amplitudes to recognize motor faults is not sufficient in a practical system. Various types of faults and load conditions will influence the spectrum structure. In order to recognize faults under various load conditions, we must consider band shift and amplitude variation as two major factors. In this paper, we use the methods of frequency axis adjustment, load interval and feature exaction to solve the band shift and amplitude variation problems respectively. After the above-mentioned procedures, efficient features are obtained. We use the Back Propagation Neural Network (BPNN) and General Regression Neural Network (GRNN) to train and recognize fault conditions.

摘要.......................... IV
Abstract.......................V
目錄...........................VII
圖目錄........................XI
表目錄........................XIII
第一章 緒論.....................1
1.1 研究背景與目的.......1
1.2 研究方法.............2
1.3 內容概述.............2
第二章 頻譜分析技術.............5
2.1 前言.....................5
2.2 傅立葉分析.....................5
2.2.1 週期性信號...............5
2.2.2 傅立葉級數...............6
2.2.3 傅立業級數的限制及傅立葉轉換........9
2.3 離散傅立葉轉換 (DFT)......................................10
2.3.1 離散傅立葉轉換(DFT)的推導.......10
2.3.2 離散傅立葉轉換(DFT)分析的限制.....11
2.3.3 數位信號處理時所使用的參數及影響....12
2.4 快速傅立葉轉換（FFT）.......................................14
2.4.1 快速傅立葉轉換(FFT)的由來.......14
第三章 研究方法....................17
3.1 類神經網路概論..........................17
3.1.1 網路之結構分類.....................21
3.1.2 網路之學習策略.....................22
3.2 倒傳遞類神經網路（BPNN）的架構及理論..23
3.2.1 BPNN的架構.......................................24
3.2.2 BPNN的學習方法....................27
3.2.3 BPNN的學習過程..............31
3.3 廣義迴歸類神經網路（GRNN）的架構及理論.33
3.3.1 GRNN簡介.......................................33
3.3.2 GRNN的理論背景....................................33
3.3.3 決定GRNN的平滑參數.................................37
3.3.4 GRNN的實行步驟..........................................40
3.3.5 GRNN與其他類神經網路之不同........................41
第四章 感應馬達故障辨識之技術..........42
4.1 前言......................42
4.2 聲音訊號的處理................................42
4.2.1 頻帶漂移...................49
4.2.1.1 波峰頻率的計算...........50
4.2.1.2 頻率軸等比校正...........52
4.2.2 振幅變化...................53
4.2.2.1 負載分段..............54
4.2.2.2 特徵萃取..............54
4.2.3 評估........................57
4.2.3.1 頻率軸等比校正的結果.........57
4.2.3.2 特徵萃取的結果.............57
第五章 感應馬達故障狀況辨識...........66
5.1 前言....................66
5.2 辨識處理程序................66
5.2.1 硬體設備架構................66
5.2.2 以聲訊為基礎之故障診斷系統.......67
5.3 參數設定.................70
5.3.1 狀況模擬...................70
5.3.2 信號處理的參數設定.............71
5.3.3 訓練樣本與辨識樣本的參數設定.........71
5.3.4 類神經網路的參數設定............74
5.4 研究方法測試結果與討論............................75
5.4.1 GRNN平滑參數 值之測試與結果.....75
5.4.2 BPNN與GRNN之測試與結果..........77
5.4.3 BPNN與GRNN故障診斷模型之架構....80
5.4.3 辨識率分析...............82
5.4.4 強韌性分析..................85
5.4.5 BPNN與GRNN特性之比較........86
第六章 結論與研究展望..............87
6.1 結論......................87
6.2 未來研究展望.....................89
參考文獻........................90

[1] B. Li and M.-Y. Chow and Y. Tipsuwan and J.C. Hung,’’ Neural-network-based motor rolling bearing fault diagnosis’’ IEEE Trans. Industrial Electronics, Volume 47, Issue 5, Oct. 2000, pp.1060 – 1069.
[2] Khuen Ho Weng and C. Yap and A. Tay and Wei Chen and Ying Zhou and W.W. Tan and Ming Chen’’ Temperature Control and In Situ Fault Detection of Wafer Warpage’’ IEEE Trans. Semiconductor Manufacturing, Volume 20, Issue 1, Feb. 2007, pp.1 – 4.
[3] T.P. Tsao,and R.C. Wu,’’The application of artifical neural network on sound signal recognition for induction motor,’’Proc. Natl. Sci. Counc.ROC(A), Vol. 23, No.1, 1999,pp. 75-84.
[4] J. W. Cooley and J.W. Tukey,“An algorithm for the machine calculation of complex Fourier series,” Mathemat1cs Computat1on, vol. 19, April 1965, pp. 297-301.
[5] R. Bracewell, The Fourier transform and its application, New York: McGraw-Hill, 1965.
[6] R.C. Singleton, “An algorithm for computing the mixed radix fast Fourier transform,” IEEE Trans. Aud1o Electroacoust., vol. AU-17, June 1969, pp. 93-103.
[7] E.O. Brigham, The fast Fourier transform, New Jersey: Prentice-Hall, 1974.
[8] A.V. Oppenheim and R.W. Schafer, Discrete-time signal processing, New Jersey: Prentice-Hall, 1989.
[9] G.D. Bergland, “A guided tour of the fast Fourier transform,” IEEE Spectrum, July 1969, pp. 41-52.
[10] P.D. Welch, “A fixed-point fast Fourier transform analysis, ”IEEE Trans. Aud1o Electroacoust., vol. 17, No. 2, 1969, pp. 151-157.
[11] 杜耿邦，“應用人工智慧於電力系統諧波源與位置之偵測”，國立中山大學電機系碩士論文，民國九十二年五月。
[12] D.F. Specht, “Probabilistic Neural Network for Classification, Mapping, or Associative Memory, ” Proc. IEEE Int. Conf. Neural Network, San Diego, CA, Vol.1, July 1988, pp. 525-532.
[13] Specht, D.F., “A General Regression Neural Network, ”IEEE Transaction on Neural Network, 1991, pp.568-576.
[14] A.J. Ellison and S.J. Yang, “Effect of rotor eccentricity on acoustic noise from induction machines,” Proc. IEE, vol. 118, No. 1, 1971, pp. 174-184.
[15] B.J. Chalmers and R. Dodgson, “Waveshapes of flux density in polyphase induction motors under saturated conditions,” IEEE Trans. Power Apparatus and Systems, vol. PAS-90, No. 2, Mar./Apr. 1971, pp. 564-569.
[16] P.J.Tavner, B. G. Gaydon, and D.M. Ward, “Monitoring generators and large motors,” IEE Proceed1ngs-B, vol. 133, No. 3, May 1986, pp. 169-180.
[17] J.R. Cameron, W.T. Thomson, and A.B. Dow, “Vibration and current monitoring for detecting airgap eccentricity in large induction motors,” IEE Proceed1ngs-B, vol. 133, No. 3, May 1986, pp. 155-163.
[18] D.R. Albright, “Intertune short-current detector for turbine-generator rotor windings, ”IEEE Trans. Power Apparatus and Systems,vol. PAS-90, No. 2, March/April 1971, pp.478-483.
[19] M. Kurtz and J.F. Lyles, “Generator insulation diagnostic testing,” IEEE Trans. Power Apparatus and Systems, vol. PAS-98, No. 5, Sept./Oct. 1979, pp. 1596-1603.
[20] R.T. Harrold, F.T. Emery, F.J. Murphy, and S.A. Drinkut, “Radio frequency sensing of incipient arcing faults within large turbine generators,” IEEE Trans. Power Apparatus and Systems, vol. PAS-98, No. 4, July/Aug. 1979, pp. 1167-1173.
[21] Y. Michiguchi, S. Tanisaka, and S. Izumi,“Development of a collector ring monitor for sparking detector of generator,” IEEE Trans. Power Apparatus and Systems, vol. PAS-102, No. 4, April 1983, pp.928-932.
[22] R. Natarajan, J.L. Kohler, and J. Sottile,“Condition monitoring of slip-ring induction motors,” Electr1c Power Systems Research, vol. 15, 1988, pp. 189-195.
[23] H.T. Yang, K.Y. Huang, and L.C. Huang, “An artificial neural network identification and control approach for the field-oriented induction motor,” Electr1c Power Systems Research, vol. 30, 1994, pp. 35-45.
[24] S. Chen, E. Zhong, and T.A. Lipo, “A new approach to motor condition monitoring in induction motor drives,” IEEE Trans. Industry Appl1cat1ons, vol. 30, No. 4, July/Aug.1994, pp. 905-911.
[25] R.R. Schoen and T.G. Habetler,“Effects of time-varying loads on rotor fault detection in induction machines.” IEEE Trans. Industry Appl1cat1ons, vol. 31, No 4, July/Aug.1995, pp. 900-906.
[26] 吳榮慶,秦純,曹大鵬,陳建源, “利用音訊於不同負載下的馬達狀況辨識,” 第十七屆電力研討會, 1996, pp. 423-427.
[27] Y.Y. Hong and Y.C. Chen “Application of algorithms and artificial-intelligence approach for locating multiple harmonics in distribution Systems ”Generation ,Transmission and Distribution, IEE Proceedings-Volume 146.
[28] 曹大鵬,吳榮慶,陳志堅,李勝義,楊裕明, “感應機聲訊狀況之研究,”第十六屆電力研討會, 1995, pp. 160-164.