本論文提出一個分析馬達電流波形並有效辨識馬達品質類別的方法，稱之為Fisher’s Linear Discriminant Analysis (Fisher’s LDA)法。本論文之Fisher’s LDA法是由下列三大單元所組成，分別為：(1)馬達信號的前置處理：包含馬達電流波形信號的擷取、雜訊的去除、信號的放大、類比/數位信號的轉換、電腦介面電路的設計等；(2)主要特徵點的選取與主要特徵值的統計：本論文以區間交集法(Range-Overlap Method，ROM)在眾多的原始特徵點中選取主要特徵點，並使用統計的方式計算主要特徵點之特徵值範圍，包含特徵值的最小值、最大值、算數平均值等；(3)辨識馬達品質的類別：本論文以Fisher’s LDA演算法辨識馬達品質的類別。依據實際的測試結果，本論文提出之Fisher’s LDA法性能如下：將「好的馬達」辨識成是「好的馬達」之平均正確率是99.92%，將「壞的馬達」正確辨識成是「壞的馬達」的正確率是92.43%，將「壞的馬達」辨識成是「好的馬達」的錯誤率是7.57%，將「好的馬達」辨識成是「壞的馬達」之錯誤率是0.08%。總平均正確辨識率為99.72%。

This study proposes a Fisher’s Linear Discriminant Analysis (Fisher’s LDA) approach to analyze current waveform for determining the motor’s quality types. Fisher’s LDA comprises three main stages: (i) the preprocessing stage for enlarging motor’s current waveforms’ amplitude and eliminating noises; (ii) the qualitative features stage for qualitative feature selection of a motor’s current waveform; (iii) the classification stage for determining motor’s quality types using the Fisher’s LDA. In the experiment, the right rate is 99.92% (92.43%) for right judgment on good (defect) motor to be determined as good (defect), the error rate is 7.57%% (0.08%) for wrong judgment on defect (good) motor to be determined as good (defect). The average right rate is 99.72%.

摘　　要 ii
Abstract iii
誌　　謝 iv
目　　錄 v
表目錄 vi
圖目錄 vii
第一章　緒論 1
1.1　前言 1
第二章　馬達信號的前置處理 3
2.1 電源自動加入裝置(APAD) 4
2.2波形偵測電路(WDC) 6
2.3增益可規劃的放大器(GPA) 6
2.4類比/數位信號轉換器(ADC) 9
第三章　特徵點的選取與特徵值的統計 10
3.1　主要的特徵點選取：區間交集法 11
3.2 主要特徵點的定義與特徵值範圍的統計 14
第四章 辨識馬達的品質類別：Fisher’s 線性鑑別分析法(FLDA)20
5.1 實驗一：測試某一個週期的馬達電流信號 25
5.2 實驗二：測試某一個馬達 27
5.3 實驗三：測試辨識準確度 29
5.4 性能比較 30
第六章 結論 31
參考文獻 32
簡 歷 36

[1] H.X. Chen, P.S.K. Chua, G.H. Lim, “Adaptive wavelet transform for vibration signal modeling and application in fault diagnosis of water hydraulic motor,” Mech. Syst. And Signal Process., vol. 20, 2006, pp. 2022-2045.
[2] S.M. Rasheed, D.W. Stashuk, M. Kamel, “An interactive environment for motor unit potential classification using certainty-based classifiers,” Simul. Model. Pract. and Theory, vol. 16, 2008, pp. 1293-1311.
[3] H. Higuchi, S.A. Endow, “Directionality and processivity of molecular motors,” Curr. Opin. in Cell Biol., vol. 14, 2002, pp. 50-57.
[4] A. Kapun, M. Curkovic, A. Hace, K. Jezernik, “Identifying dynamic model parameters of a BLDC motor,” Simul. Model. Pract. and Theory, vol. 16, 2008, pp. 1254-1265.
[5] M. Ahdesmäki, K. Strimmer, “Feature selection in omics prediction problems using cat scores and false nondiscovery rate control,” Ann. of Appl. Statistics, vol. 4, 2010, pp. 503–519
[6] H. Yu, J. Yang, “A direct LDA algorithm for high-dimensional data with application to face recognition,” Pattern Recog., vol. 34, 2001, pp. 2067–2069
[7] G. Perriere, J. Thioulouse, “Use of correspondence discriminant analysis to predict the subcellular location of bacterial proteins,” Comput. Methods and Programs in Biomed, vol. 70, 2003, pp. 99–105.
[8] T. C. Chiang, C. H. Chiu, W.J. Wang, “A fuzzy rule based adaptive centre weighted median filter, “ Int. J. Comput. Appl. in Technol., vol. 27, 2006, pp. 192-203.
[9]. Y. C. Yeh, “Fuzzy Logic Method for Motor Quality Types on Current Waveforms, “ Measurement, vol. 46, 2013, pp. 1682-1691.
[10] S. Weber, “Electronic Circuits Notebook: Proven Designs for Systems Applications,” Published by McGraw-Hill Inc.,US, 1981.
[11] http://www.alldatasheet.com/view.jsp?Searchword=MP7523.
[12] 林綠綺,葉雲奇,楚萃瑤, “使用群聚分析法辨識馬達的品質類別,” Proceedings of 2013 National Symposium on Systems Science and Engineering, pp. 18-20, 2013.
[13] 黃煌翔, “介面技術與週邊設備,”全華圖書.
[14] A. Jain, D. Zongker, “Feature selection: Evaluation, application, and small sample performance,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 19, 1997, pp. 153-158.
[15] M. Dash, H. Liu, “Feature selection for classification,” Intelligent Data Analysis, vol. 1, 1997, pp. 131-156
[16] T. S. Lin, J. Meador, “Statistical feature extraction and selection for IC test pattern analysis,” Proc. Circuits and systems, vol. 1, 1992, pp. 391-394
[17] P. J. G. Lisboa, R. Mehri-Dehnavi, “Sensitivity methods for variable selection using the MLP,” International Workshop on Neural Networks for Identification, Control, Robotics and Signal/Image, 1996, pp. 330-338.
[18] R. Biswas, P. Goel, A. Mukerjee, H. Shawky, “The application of genetic algorithms for a three-way optimization of risk-return-tax tradeoff in equity mutual fund portfolios,” University at Albany working paper, 2004.
[19] A. Prinzie, D. Van den Poel, “Random Forests for multiclass classification: Random Multi-Nomial Logit,” Expert Systems with Applications, vol. 34, 2008, pp. 1721–1732.
[20] Y. C. Yeh, W. J. Wang, C.W. Chiou, “Feature selection algorithm for ECG signals using Range-Overlap Method,” Expert System with Application, vol. 37, 2009, pp. 3499-3512.
[21] Y. C. Yeh, W. J. Wang, C.W. Chiou, “Cardiac Arrhythmia Diagnosis Method using Linear Discriminant Analysis on ECG Signals,” Measurement, vol. 42, 2009, pp. 778-789
[22] Y. C. Yeh, Y. Chu, C. W. Chiou, H. J. Lin, “A Fuzzy C-Means Method for Determining Motor’s Quality Types Based on Current Waveforms,” Lecture Notes in Electrical Engineering, vol. 234, 2013, pp. 77-83.
[23] L. C. Lin, Y. C. Yeh, “Effective Motor’s Quality Types Determination on Motor’s Current Waveforms Using the Euclidean Distance Measurement Method,” International Symposium on Computer, Consumer and Control, Taichung, Taiwan, June 10-12,