本論文提出一個有效的馬達診斷方法，可以在適應大範圍的轉速情況下，診斷不同故障種類、不同故障程度的永磁同步馬達。人工神經網路已經被廣泛的應用於此種馬達診斷的領域中。本論文採用馬達定子電流作為診斷的參考訊號。本論文的馬達診斷系統，在經過特徵提取與特徵分類後將會被完成。
本研究中，我們提出了兩種特徵提取的方法：第一種是深度學習架構的1D卷積神經網路(1D convolutional neural networks, 1D-CNN)，第二種是深度學習架構的堆疊自編碼器(stacked autoencoder, SAE)。我們同時也將我們的深度學習架構，與傳統的訊號分析方法，小波包轉換(wavelet packet transform, WPT)相互比較。深度學習與小波包轉換的主要差別在於，深度學習不需要手動調整特徵提取的參數，但小波包轉換需要調整帶通濾波器的參數，甚是手動提取特徵。
本實驗使用真實馬達數據進行實驗，實驗結果表明所提出的實時馬達狀態監測方法的有效性。實驗結果也表明，該方法可以有效的診斷馬達的五種不同的狀態，其中有些狀態是不同情況，有些狀態則是在相同的故障情況但不同程度。馬達狀態包含正常馬達狀態，兩種不同程度的退磁故障，以及兩種軸承故障。

This manuscript presents an effective diagnosis algorithm for permanent magnet synchronous motors running with an array of faults of varying severity over a wide speed range. Artificial neural networks are widely used for diagnosing such conditions. The fault diagnosis is based on current signature analysis. The complete fault motor diagnosis system requires the extraction of features based on the current method and a subsequent method for adding classifications.
In this study, we propose two feature extraction methods: the first is a deep one-dimensional convolution neural network (1D-CNN) that includes a softmax layer.; the second is also a deep neural network, stacked autoencoder (SAE). We also compared our methods with a traditional algorithm, wavelet package transform (WPT). The main difference between our methods and WPT is that out methods do not require manual parameter adjustment, whereas the WPT needs to adjust the parameters of the band-pass filters.
Experimental results obtained using real-time motor stator current data demonstrate the effectiveness of the proposed methods for real-time monitoring of motor conditions. The results also show that the proposed methods can effectively diagnose five different motor states, including two different demagnetization fault states and two bearing fault states.

論文審定書 i
致　謝 ii
摘　要 iii
Abstract iv
目 錄 v
圖 次 ix
表 次 xiii
第一章 緒論 1
1-1 前言 1
1-2 文獻回顧 2
1-3 研究動機與範疇 7
1-4 研究主要貢獻 9
1-5 組織章節 10
第二章 達定子電流訊號觀察 12
2-1 前言 12
2-2 永磁同步馬達 13
2-2-1 基本介紹……………………………………………….. 13
2-2-2 馬達規格……………………………………………….. 13
2-2-3 電氣角頻率與機械角頻率關係……………………….. 15
2-3 正常馬達定子電流訊號 16
2-4 退磁故障馬達定子電流訊號 17
2-4-1 退磁故障現象與原因………………………………….. 17
2-4-2 退磁故障馬達的製造………………………………….. 18
2-4-3 退磁故障馬達的時頻特徵…………………………….. 18
2-5 軸承故障馬達定子電流訊號 21
2-5-1 軸承故障現象與原因………………………………….. 21
2-5-2 軸承故障現象與原因………………………………….. 22
2-5-3 軸承故障時頻現象…………………………………….. 22
第三章 深度學習架構 26
3-1 前言 26
3-2 人工神經網路 26
3-2-1 人工神經網路基本介紹……………………………….. 26
3-2-2 反向傳播……………………………………………….. 28
3-2-3 損失函數與優化器…………………………………….. 30
3-2-4 深度人工神經網路…………………………………….. 33
3-3 堆疊自編碼器 34
3-3-1 自編碼器基本介紹…………………………………….. 34
3-3-2 本論文的堆疊自編碼器架構………………………….. 36
3-4 卷積神經網路 37
3-4-1 卷積神經網路簡介…………………………………….. 37
3-4-2 局部感知機…………………………………………….. 38
3-4-3 權值共享……………………………………………….. 39
3-4-4 池化…………………………………………………….. 40
3-4-5 卷積神經網路基本架構……………………………….. 41
3-4-6 本論文的卷積神經網路架構………………………….. 41
3-5 Softmax感知機 45
第四章 實驗結果 47
4-1 實驗架構 47
4-1-1 系統架構……………………………………………….. 47
4-1-2 硬體架構……………………………………………….. 48
4-1-3 馬達定子電流訊號收集……………………………….. 51
4-1-4 神經網路訓練………………………………………….. 52
4-2 深度學習分類正確率結果 52
4-3 診斷結果分析 56
4-4 研究貢獻與比較 58
4-4 診斷電腦規格 59
第五章 結論與未來展望 61
5-1 結果討論 61
5-2 未來展望 62
參考文獻 64
簡 歷 71

[1] 朱效賢、潘敏俊, “包絡譜分析於軸承故障診斷之探討暨工程應用”，碩士論文，國立中央大學機械工程研究所，民國九十四年九月.
[2] 康淵、趙安民, “馬達故障診斷之模糊類神經網路”，碩士論文，國立中央大學機械工程研究所，民國九十三年七月.
[3] X. Dai and Z. Gao, “From Model, Signal to Knowledge: A Data-Driven Perspective of Fault Detection and Diagnosis,” IEEE Trans. Ind. Informat., vol. 9, no. 4, pp. 2226-2238, Nov. 2013.
[4] A. Ahmad and L. Dey, “A k-mean clustering algorithm for mixed numeric and categorical data,” Data & Knowledge Engineering, vol. 63, no. 2, pp. 503–527, Nov. 2007.
[5] W. Zhou et al., “ISAR imaging of targets with rotating parts based on robust principal component analysis,” IET Radar Sonar Navigat., vol. 11, no. 4, pp. 563–569, Apr. 2017.
[6] A. Maligo and S. Lacroix, “Classification of outdoor 3D lidar data based on unsupervised Gaussian mixture models,” IEEE Trans. Autom. Sci. Eng., vol. 14, no. 1, pp. 5–16, Jan. 2016.
[7] M.N. Bawane and K.M. Bhurchandi. “Classification of mental task based on EEG processing using self organising feature map,” in IEEE Int. Conf. Intell. Human Mach. Syst. Cybern., Hangzhou, China, 2014, pp. 240–244.
[8] Z. Szadkowski et al., “Optimization of an FPGA trigger based on an artificial neural network for the detection of neutrino-induced air showers,” IEEE Trans. Nucl. Sci., vol. 64, no. 6, pp. 1271–1281, Jan. 2017.
[9] S.U. Jan et al., “Sensor fault classification based on support vector machine and statistical time-domain features,” IEEE Access., vol. 5, pp. 8682–8690, May. 2017.
[10] F.Y. Wu. and H.H. Asada, “Implicit and intuitive grasp posture control for wearable robotic fingers: a data-driven method using partial least squares,” IEEE Trans. Robot., vol. 32, no. 1, pp. 176–186, Feb. 2016.
[11] J. Lin and L. Qu, “Feature extraction based on morlet wavelet and its application for mechanical fault diagnosis,” J. Sound. Vib., vol. 234, no. 1, pp. 135–148, Jun. 2000.
[12] M.A.S.K. Khan and M.A. Rahman, “Development and implementation of a novel fault diagnostic and protection technique for IPM motor drives,” IEEE Trans. Ind. Electron., vol. 56, no. 1, pp. 85–92, Jan. 2009.
[13] E. G. Strangas, S. Aviyente, and S. S. H. Zaidi, “Time-frequency analysis for efficient fault diagnosis and failure prognosis for interior permanent-magnet AC motors,” IEEE Trans. Ind. Electron., vol. 55, no.12, pp. 4191-4199, Dec. 2008.
[14] P.K. Kankar, S.C. Sharma, and S.P. Harsha, “Fault diagnosis of ball bearings using machine learning methods,” Int. J. Expert. Syst. Appl., vol. 38, no. 3, pp.1876–1886, Mar. 2011.
[15] B. Li et al., “Neural-network-based motor rolling bearing fault diagnosis,” IEEE Trans. Ind. Electron., vol. 47, no. 5, pp. 1060–1069, Oct. 2000.
[16] V. Sugumaran, and K.I. Ramachandran, “Automatic rule learning using decision tree for fuzzy classifier in fault diagnosis of roller bearing,” Mech. Syst. Signal. Process., vol. 21, no. 5, pp. 2237–2247, Jul. 2007.
[17] B. Samanta and K.R. Al-Balushi, “Artificial neural network based fault diagnostics of rolling element bearings using time-domain features,” Mech. Syst. Signal. Process, vol. 17, no. 2, pp. 317-328, Mar. 2003.
[18] B. Samanta, K.R. Al-Balushi, and S.A. Al-Araimi, “Artificial neural networks and support vector machines with genetic algorithm for bearing fault detection,” Eng. Appl. Artif. Intell., vol. 16, no. 7-8, pp. 657-665, Oct-Dec. 2003.
[19] C.J.C. Burges, “A tutorial on support vector machines for pattern recognition,” J. Data. Min. Knowl. Discov., vol. 2, no. 2, pp. 121-167, Jun. 1998.
[20] H. Shao et al., “A novel deep autoencoder feature learning method for rotating machinery fault diagnosis,” Mech. Syst. Signal. Process., vol. 95, pp. 187–204, Oct. 2017.
[21] P. Tamilselvan and P. Wang, “Failure diagnosis using deep belief learning based health state classification,” Reliab. Eng. Syst. Saf., vol. 115, pp. 124–135, Jul. 2013.
[22] W. Sun et al., “Convolutional discriminative feature learning for induction motor fault diagnosis,” IEEE Trans. Ind. Informat., vol. 13, no. 3, pp. 1350–1359, Jun. 2017.
[23] T. Ince et al., “Real-time motor fault detection by 1-D convolutional neural networks,” IEEE Trans. Ind. Electron., vol. 63, no. 11, pp. 7067–7075, Nov. 2016.
[24] 劉昌換、吳添保、黃仲欽、彭榮芳、楊宗銘、詹前茂、劉添華、賴炎生, “電機機械”，東華書局，明國九十一年.
[25] 王威仁、彭昭暐, “基於小波轉換與時頻分析之永磁同步馬達故障診斷”，碩士論文，國立中山大學機械與機電工程研究所，民國一百零五年七月.
[26] 王建昌, “高速永磁同步馬達驅動技術”，財團法人工業技術研究院，能源知識庫，民國一百零二年.
[27] Y. Liang, “Diagnosis of inter-turn short-circuit stator winding fault in PMSM based on stator current and noise,” in IEEE Int. Conf. Ind. Tech., Busan, South Korea, 2014, pp. 138-142.
[28] M.E.H. Benbouzid, “A review of induction motors signature analysis as a medium for faults detection,” IEEE Trans. Ind. Electron., vol. 47, no. 5, pp. 984–993, Oct. 2000.
[29] J. R. Stack, R. G. Harley, and T. G. Habetler,“An amplitude modulation detector for fault diagnosis in rolling element bearings,”IEEE Trans. Ind. Electron., vol. 51, no. 5, pp. 1097-1102, 2004.
[30] 張宏展、鄭強、鄭育明、郭政謙、黃立彬, “馬達運轉頻譜監測系統之研製”，中華民國第三十五屆電力工程研討會，台灣科技大學電機工程系，民國一百零三年.
[31] J. Schmidhuber, “Deep learning in neural networks: An overview,” Neural Network, vol. 8, pp. 85-117, Oct. 2014.
[32] Y. LeCun, Y. Bengio, and G. Hinton, “Deep learning,” Nature, vol. 521, no. 7553, pp. 436-444, May. 2015.
[33] Y. Bengio, A. Courville, and P. Vincent, “Representation learning: A review and new perspectives,” IEEE Trans. Pattern. Anal. Mach. Intell., vol. 35, no. 8, pp. 1798-1828, Mar. 2013.
[34] B.A. Olshausen and D.J. Field, “Emergence of simple-cell receptive field properties by learning a sparse code for natural images,” Nature, vol. 381, no. 6583, pp. 607-609, Jun. 1996.
[35] D.E. Rumelhart, G.E. Hinton, and R.J. Williams, “Learning representations by back-propagating errors,” Nature, vol. 323, no. 6088, pp. 533-536, Oct. 1986.
[36] R.A. Jacobs, “Increased rates of convergence through learning rate adaptation,” Neural Network, vol. 1, no. 4, pp. 295-307, Mar. 1988.
[37] D. Kingma and J. Ba, “Adam: A method for stochastic optimization,” in 3rd Int. Conf. Learn. Representations, San Diego, CA, USA, 2015, pp. 1–15.
[38] H.W. Lin, M. Tegmark, and D. Rolnick, “Why does deep and cheap learning work so well,” J. Stat. Phys., Jul. 2017.
[39] R. Meyer, “Deep learning smarts up your smart phone,” AMAX, Dec. 2015.
[40] C.Y. Liou, J.C. Huang, and W.C. Yang, “Modeling word perception using the Elman network,” Neurocomputing, vol. 71, no. 16, pp. 3150-3157, Jun. 2008.
[41] C.Y. Liou et al., “Autoencoder for words,” Neurocomputing, vol. 139, no. 2, pp. 84-96, Sep. 2014.
[42] Diederik P. Kingma and M. Welling, “Auto-Encoding Variational Bayes,” [Online]. Available:http://arXiv:1312.6114.
[43] M.B. Lazreg, M. Goodwin, and O.C. Granmo., “Vector representation of non-standard spellings using dynamic time warping and a denoising autoencoder,” in IEEE Congress on Evolutionary Computation, Donostia, San Sebastián, Spain, Spain, 2017, pp. 1444–1450.
[44] F. Lv, M. Han and T. Qiu, “Remote sensing image classification based on ensemble extreme learning machine with stacked autoencoder,” IEEE Access, vol 5, pp. 9021–9031, May. 2017.
[45] Z. Chen and W. Li, “Multisensor feature fusion for bearing fault diagnosis using sparse autoencoder and deep belief network,” IEEE Trans. Instrum. Meas., vol. 66, no. 7, pp. 1693–1702, Jul. 2017.
[46] D. H. Hubel and T. N. Wiesel, “Receptive fields, binocular interaction and functional architecture in the cat's visual cortex,” J. Physiol., vol. 160, no. 1, pp. 106-154, Jan. 1962.
[47] K. Fukushima and S. Miyake, “Neocognitron: A Self-Organizing Neural Network Model for a Mechanism of Visual Pattern Recognition,” in Competition and Cooperation in Neural Nets: Proceedings of the U.S.-Japan Joint Seminar held at Kyoto, Japan February 15–19, 1982, Springer Berlin Heidelberg: Berlin, Heidelberg., S. I. Amari and M. A. Arbib, Editors. 1982, pp. 267-285.
[48] G. Liu et al., “Passenger flow estimation based on convolutional neural network in public transportation system,” Knowl. Base. Syst., vol. 123, pp. 102–115, May. 2017.
[49] J. Deng et al., “ImageNet: a large-scale hierarchical image database,” in IEEE Conf. Comput. Vis. Pattern Recognition, Miami, FL, USA, 2009, pp. 248–255.
[50] G. E. Hinton et al., “Improving neural networks by preventing co-adaptation of feature detectors,” [Online]. Available: http://arxiv.org/abs/1207.580.
[51] N. Srivastava et al., “Dropout: A simple way to prevent neural networks from overfitting,” J. Mach. Learn. Res., vol. 15, no. 1, pp. 1929-1958, Jun. 2014.
[52] B.M. Ebrahimi et al., “Advanced eccentricity fault recognition in permanent magnet synchronous motors using stator current signature analysis,” IEEE Trans. Ind. Electron., vol. 61, no. 4, pp. 2041-2052, Apr. 2014.