機械振動對於工具機是不可避免的，現今已有許多針對振動的分析，例如切削顫振偵測、故障診斷等。
本論文主要針對小型五軸加工機振動訊號的研究，嘗試建立一種可運用於工具機的健康診斷法。使用三台小型五軸加工機以相同參數及運動程式進行實驗，搭配智慧預測與診斷系統(Prediction Diagnosis Performance System) 收集振動訊號，再使用軟體處理訊號與主成分分析。將振動訊號分析提取6個時域特徵(Root Mean Square均方根值、Kurtosis峰度、Variance變異數、Crest Factor 峰值因數、Standard Deviation 標準差、Skewness 偏態係數)、10個頻域特徵(將每2560Hz頻寬切割成10段，每段256Hz取頻率平均值作為頻域特徵)、10個包絡線頻譜特徵(將2560Hz頻寬切割成10段，每256Hz裡不同頻率振幅的最高點連成一曲線即為包絡線，再將每段萃取一個最高頻率作為特徵)、8個功率譜密度特徵(Power Spectral Density，為訊號的平均功率在每一頻率的密度)、1個轉速頻特徵(計算設定轉速的頻率，提取作為特徵)共35個特徵，透過主成分分析(Principal Component Analysis)的統計原理，選取出對資料影響最大的主要特徵，將資料投影成分布圖。
實驗中收集正常與異常偏心數據進行比較，重複收集三組數據確保資料重複性，結果顯示以小型五軸加工機為載具，使用主成分分析結果來判斷機台健康狀態是可行的。

In this study, principal component analysis (PCA) was used to establish a dynamic performance diagnosis method for machine tools. The health of the machine was determined through the processing and analysis of vibration, current, and temperature signals obtained from the PCA analysis. Further, we conducted an online diagnosis for machine performance. Three small five-axis processing machines were used for the experimental study with the same parameters and motion programs, and the prediction diagnosis performance system (PDPS) was used to collect, analyze, and process the dynamic signals of the machine. The health data and eccentricity of the spindle were collected as abnormal comparison data. The experiment was repeated to collect three normal and abnormal states, and multiple sets of data were compared. The results show that the data for dynamic feature diagnosis using this method is regular, thus it can be used to construct a health feature model to digitally compare and monitor the health of the machine. The experimental results were analyzed and compared using the PCA projection data point distribution map. This method can also be used to compare the characteristics of the three machines and the differences in the normal and abnormal data distribution. Thus, it can be applied to related vehicles as a performance indicator for machine detection. The test and analysis results indicated that the vibration, current, and temperature signals obtained by the PCA were concentrated and regular. When the machine was abnormal, the data distribution deviated from the distribution range of the health characteristics.

摘要.....i
Abstract.....ii
誌謝.....iii
目錄.....iv
表目錄.....vi
圖目錄.....vii
符號說明.....x
第一章 緒論.....1
1.1研究背景.....1
1.2研究目的.....1
1.3文獻探討.....1
1.4論文架構.....3
第二章 實驗基礎理論介紹.....6
2.1振動的基礎理論.....6
2.1.1單自由度系統.....6
2.2加速規基本原理.....7
2.3振動訊號分析重要名詞.....8
2.4濾離群值.....11
2.4.1卡爾曼濾波（Kalman Filter）.....11
2.4.2 Z - Score.....13
2.5主成分分析(Principal Components Analysis,PCA).....14
2.6高斯混合模型(Gaussian Mixture Model,GMM).....15
第三章 實驗設備與軟體介紹.....17
3.1 PMC小型五軸加工機.....17
3.2 智慧預測與診斷系統(Prediction Diagnosis Performance System).....18
3.3 加速規.....19
3.4 溫度感測器.....21
3.5 電流感測器.....22
3.6 主軸產生偏心治具.....23
3.7 訊號擷取與即時監測軟體.....23
3.8 訊號分析軟體.....25
第四章 實驗規劃及介紹.....26
4.1 實驗設備安裝介紹.....26
4.1.1 主軸偏心治具安裝.....27
4.2 振動訊號擷取.....28
4.3 實驗參數及運動程式制定.....28
4.3.1 基本參數選定.....28
4.3.2 運動程式制定.....29
4.4 數據收集及分析.....31
4.4.1 振動數據提取.....32
4.4.2 振動訊號檢查.....33
4.4.3 振動訊號處理.....34
4.4.4 特徵萃取.....34
4.4.5 主成分分析(Principal Components Analysis).....36
第五章 實驗結果與討論.....37
5.1振動訊號RMS分析比較.....37
5.2整機振動訊號主成分分析結果.....42
5.2.1主軸振動訊號主成分分析結果.....46
5.3多機健康診斷分析結果比較.....50
5.3.1多機主軸分析結果比較.....52
5.4多種異常狀態模擬分析結果比較.....54
5.4.1各異常模擬分析結合比較.....60
5.5建立健康診斷模型及模擬診斷.....61
第六章 結論與未來展望.....64
6.1結論.....64
6.2未來展望.....64
參考文獻.....65
Extended Abstract.....67
Abstract.....67
1.Introduction.....68
2.Experimental equipment.....68
3.Vibration signal capture.....68
4.Vibration Analysis and Comparison.....69
5.Discussions.....71
6.Conclusion.....71
Reference.....71

[1] Zhiqiang Wang, Mathieu Ritou, Catherine Da Cunha, Benoît Furet, 2020, “Contextual classification for smart machining based on unsupervised machine learning by Gaussian
Mixture model,” International Journal of Computer Integrated Manufacturing, Vol. 33, pp. 1042-1054.
[2] Hong Seok Park, Bowen Qi, Duck Viet Dang, Dae Yu Park, 2018, “Development of smart machining system for optimizing feedrates to minimize machining time,” Journal of Computational Design and Engineering, Vol. 5, pp. 299-304.
[3] Qiulian Wang, Hui Yang, 2020, “Sensor-Based Recurrence Analysis of Energy Efficiency in Machining Processes,” IEEE Access, Vol. 8, pp. 18326-18336.
[4] Tzu Chi Chan, Hsin Hsien Lin, Yu Chuan Wang, Chang Chia Chuan, Han Huel Lin, 2020, “Intelligent Diagnosis for Cutting Processes,” IEEE Eurasia Conference on IOT, Communication and Engineering (ECICE), pp. 140-143, Yunlin, Taiwan.
[5] Xiaochen Zhang, Dongxiang Jiang, Te Han, Nanfei Wang, Wenguang Yang, Yizhou Yang,2017, “Rotating Machinery Fault Diagnosis for Imbalanced Data Based on Fast Clustering Algorithm and Support Vector Machine,” Journal of Sensors, Vol. 2017, pp. 15.
[6] Mohamed Elforjani, Suliman Shanb, 2018, “Prognosis of Bearing Acoustic Emission Signals Using Supervised Machine Learning,” IEEE Transactions on Industrial Electronics, Vol. 65, pp. 5864-5871.
[7] Deepam Goyal, S. S. Dhami, B. S. Pabla, 2020, “Non-Contact Fault Diagnosis of Bearings in Machine Learning Environment,” IEEE Sensors Journal, Vol. 20, pp. 4816-4823.
[8] Abdenour Soualhi, Kamal Medjaher, Noureddine Zerhouni, 2015, “Bearing Health Monitoring Based on Hilbert–Huang Transform, Support Vector Machine, and Regression,” IEEE Transactions on Instrumentation and Measurement, Vol. 64, pp. 52-62.
[9] Linbo Zhu, Dong Chen, Pengfei Feng, 2021, “Equipment Operational Reliability Evaluation Method Based on RVM and PCA-Fused Features,” Mathematical Problems in Engineering, Vol. 2021, pp. 9.
[10] Alessandra Caggiano, 2018, “Tool Wear Prediction in Ti-6Al-4V Machining through Multiple Sensor Monitoring and PCA Features Pattern Recognition,” Sensors (Basel, Switzerland), Vol. 18, Issue. 3.
[11] Shaojiang Dong, Dihua Sun, Baoping Tang, Zhenyuan Gao, Wentao Yu, Ming Xia, 2014, “A Fault Diagnosis Method for Rotating Machinery Based on PCA and Morlet Kernel SVM,” Mathematical Problems in Engineering, Vol. 2014, pp. 8.
[12] Jianjian Yuan, HuaShao, Yun Cai, Xinhua Shi, 2021, “Energy efficiency state identification of milling processing based on EEMD-PCA-ICA,” Measurement, Vol. 174.
[13] Zhou Yuqing, Sun Bingtao, Li Fengping, Song Wenlei, 2015, “NC Machine Tools Fault Diagnosis Based on Kernel PCA and k-Nearest Neighbor Using Vibration Signals,” Shock and Vibration, Vol. 2015, pp. 10.
[14] Ziyuan Jiang, Qinkai Han, Xueping Xu, 2020 “Fault Diagnosis of Planetary Gearbox Based on Motor Current Signal Analysis,” Shock and Vibration, Vol. 2020, pp. 13.
[15] Fei Chen, Yifeng Ye, Weizheng Chen, Binbin Xu, Chao Chen, Zhaojun Yang, 2017, “Fault diagnosis of motorized spindle via modified empirical wavelet transform-kernel PCA and optimized support vector machine,” Journal of Vibroengineering, Vol. 19, Issue 4, pp. 2611-2631.
[16] Huaitao Shi, Jin Guo, Xiaotian Bai, Lei Guo, Zhenpeng Liu, Jie Sun, 2020, “Gearbox Incipient Fault Detection Based on Deep Recursive Dynamic Principal Component Analysis,” IEEE Access, Vol. 8, pp. 57646-57660.
[17] Moussa Hamadache, Dongik Lee, Kalyana C. Veluvolu, 2015, “Rotor Speed-Based Bearing Fault Diagnosis (RSB-BFD) Under Variable Speed and Constant Load,” IEEE Transactions on Industrial Electronics, Vol. 62, pp. 6486-6495.
[18] Miguel Marques, João Martins, V. Fernão Pires, Rui Dias Jorge, Luís Filipe Mendes, 2013, “Fault Detection and Diagnosis in Induction Machines: A Case Study,” Camarinha-Matos L.M., Tomic S., Graça P. (eds) Technological Innovation for the Internet of Things, Vol. 394.
[19] Silvia M. Zanoli, Giacomo Astolfi, “Application of a Fault Detection and Isolation System on a Rotary Machine,” International Journal of Rotating Machinery, Vol. 2013, pp. 11.
[20] 伍尚宏，2019，「五軸工具機結構分析與智慧化加工」，國立虎尾科技大學機械與電腦輔助工程系，碩士論文。
[21] National Instruments，2020， “Measuring Vibration with Accelerometers” ，加速規作用原理參考。