在現代工業高速發展的基礎下，如何妥善的達成預測性維護已成為相當重要的課題。該策略透過預先偵測設備的訊號異常徵兆，預測出關鍵零件的剩餘使用壽命 (RUL) 並發出警示，也可以同時幫助安排維修任務和庫存的供給。應用不同方法對軸承的震動訊號做分類及預測的問題，已有不少研究著墨，然而利用變分自動編碼器 (VAE) 作為混合特徵的結構，可視為一種未被考慮過的新觀點。本研究提出了一個三階段的方法，結合了特徵混合，異常偵測和深度學習預測模型。除了使用 VAE 混合特徵外，指數加權移動平均 (EWMA) 和六標準差 (6σ) 管制圖被用於進行異常點偵測，幫助我們計算從異常發生到零件生命結束 (EOL) 的相關參數，且被用於從訓練資料集到測試資料集的遷移學習。接著建構深度學習模型，如具注意力模型的雙向長短期記憶模型 (BiLSTM with attention) 來做 RUL 預測。本研究使用 IEEE PHM 2012 Data Challenge 官方提供的數據集，將三個不同條件下的軸承資料各取一個作為訓練資料集，剩餘十一個軸承做為測試集。評估標準參考競賽提供的 Performance score，該分數對預先預測結果有利，結果表明在比較的五個方法中，提出方法屬於具競爭力的高分。

With the rapid development of modern industry, how to properly achieve predictive maintenance strategy has become a significant issue. The strategy predicts the remaining useful life (RUL) of critical parts by detecting abnormal signals of equipment in advance, and can also help arrange maintenance tasks and inventory supply at the same time. This thesis proposes a three-stage approach that combines feature fusion, anomaly detection, and a deep learning predictive model. The problem of classifying and predicting the vibration signals of bearings using different methods had been studied a lot. However, taking variational autoencoders (VAE) as the structure of feature fusion can be regarded as a new point of view that has not been considered. In addition to using VAE fusing features, exponentially weighted moving average (EWMA) and six sigma (6σ) control charts are used for anomaly detection, helping us calculate relevant parameters from anomaly occurrence to end of life (EOL), applied for the transfer learning from training dataset to testing dataset. Then construct a deep learning model, such as a bidirectional long short-term memory model with attention model (BiLSTM with attention) for RUL prediction. This thesis uses the dataset officially released the IEEE PHM 2012 Prognostic Challenge, taking one bearing data under three conditions as the training dataset, and the remaining eleven bearings are used as the testing dataset. The evaluation criteria refer to the performance score provided by the competition, which is beneficial to the advance predict results. Finally, the comparison results show that the proposed method has a highly competitive score compared with the previous studies.

[1] Y. Wang, B. Tang, Y. Qin, and T. Huang, “Rolling bearing fault detection of civil aircraft
engine based on adaptive estimation of instantaneous angular speed,” IEEE Trans. Ind.
Informatics, vol. 16, no. 7, pp. 4938–4948, Jul. 2020, doi: 10.1109/TII.2019.2949000.
[2] H. Li, T. Liu, X. Wu, and Q. Chen, “Research on bearing fault feature extraction based on
singular value decomposition and optimized frequency band entropy,” Mech. Syst. Signal
Process., vol. 118, pp. 477–502, Mar. 2019, doi: 10.1016/j.ymssp.2018.08.056.
[3] P. F. Albrecht et al., “Assessment of the reliability of motors in utility applications,” IEEE
Power Eng. Rev., vol. PER-7, no. 9, pp. 41–41, Sep. 1987, doi:
10.1109/MPER.1987.5527415.
[4] D. Z. and P. J. T. C. J. Crabtree, “Survey of commercially available condition monitoring
systems for wind turbines,” Durham Univ. Sch. Eng. Comput. Sci , May 2014, Accessed: Apr.
10, 2022. [Online]. Available: http://dro.dur.ac.uk/12497/
[5] A. K. S. Jardine, D. Lin, and D. Banjevic, “A review on machinery diagnostics and
prognostics implementing condition-based maintenance,” Mech. Syst. Signal Process., vol. 20,
no. 7, pp. 1483–1510, Oct. 2006, doi: 10.1016/j.ymssp.2005.09.012.
[6] I. H. Afefy, “Reliability-centered maintenance methodology and application: a case study,”
Engineering, vol. 02, no. 11, pp. 863–873, 2010, doi: 10.4236/eng.2010.211109.
[7] J. Lee, F. Wu, W. Zhao, M. Ghaffari, L. Liao, and D. Siegel, “Prognostics and health
management design for rotary machinery systems—Reviews, methodology and applications,”
Mech. Syst. Signal Process., vol. 42, no. 1–2, pp. 314–334, Jan. 2014, doi:
10.1016/j.ymssp.2013.06.004.
[8] R. Yan, X. Chen, W. Li, and S. Sheng, “Mathematical methods and modeling in machine fault
diagnosis,” Math. Probl. Eng., vol. 2014, pp. 1–3, 2014, doi: 10.1155/2014/516590.
38
[9] C. Xing, L. Ma, and X. Yang, “Stacked denoise autoencoder based feature extraction and
classification for hyperspectral images,” J. Sensors, vol. 2016, pp. 1–10, 2016, doi:
10.1155/2016/3632943.
[10] A. Vaswani et al., “Attention is all you need,” Adv. Neural Inf. Process. Syst., Jun. 2017, doi:
10.48550/arXiv.1706.03762.
[11] M. Jouin, R. Gouriveau, D. Hissel, M.-C. Péra, and N. Zerhouni, “Particle filter-based
prognostics: review, discussion and perspectives,” Mech. Syst. Signal Process., vol. 72–73, pp.
2–31, May 2016, doi: 10.1016/j.ymssp.2015.11.008.
[12] P. Nectoux, R. Gouriveau, K. Medjaher, E. Ramasso, and M. Chebel, “Pronostia : an
experimental platform for bearings accelerated degradation tests,” Proc. IEEE Int. Conf. Prog.
Heal. Manag., pp. 1–6, Jul. 2012, Accessed: Apr. 10, 2022. [Online]. Available:
https://hal.archives-ouvertes.fr/hal-00719503
[13] Y. Lei, N. Li, S. Gontarz, J. Lin, S. Radkowski, and J. Dybala, “A model-based method for
remaining useful life prediction of machinery,” IEEE Trans. Reliab., vol. 65, no. 3, pp. 1314–
1326, Sep. 2016, doi: 10.1109/TR.2016.2570568.
[14] Z. Zhang, X. Si, C. Hu, and Y. Lei, “Degradation data analysis and remaining useful life
estimation: a review on Wiener-process-based methods,” Eur. J. Oper. Res., vol. 271, no. 3,
pp. 775–796, Dec. 2018, doi: 10.1016/j.ejor.2018.02.033.
[15] Y. Wang, Y. Zhao, and S. Addepalli, “Remaining useful life prediction using deep learning
approaches: a review,” Procedia Manuf., vol. 49, pp. 81–88, 2020, doi:
10.1016/j.promfg.2020.06.015.
[16] E. Sutrisno, H. Oh, A. S. S. Vasan, and M. Pecht, “Estimation of remaining useful life of ball
bearings using data driven methodologies,” in 2012 IEEE Conference on Prognostics and
Health Management, Jun. 2012, pp. 1–7. doi: 10.1109/ICPHM.2012.6299548.
[17] H. Qiu, J. Lee, J. Lin, and G. Yu, “Robust performance degradation assessment methods for
enhanced rolling element bearing prognostics,” Adv. Eng. Informatics, vol. 17, no. 3–4, pp.
39
127–140, Jul. 2003, doi: 10.1016/j.aei.2004.08.001.
[18] S. Hong, Z. Zhou, E. Zio, and K. Hong, “Condition assessment for the performance
degradation of bearing based on a combinatorial feature extraction method,” Digit. Signal
Process., vol. 27, pp. 159–166, Apr. 2014, doi: 10.1016/j.dsp.2013.12.010.
[19] Y. Xu, Z. Li, S. Wang, W. Li, T. Sarkodie-Gyan, and S. Feng, “A hybrid deep-learning model
for fault diagnosis of rolling bearings,” Measurement, vol. 169, p. 108502, Feb. 2021, doi:
10.1016/j.measurement.2020.108502.
[20] C. Hu, B. D. Youn, P. Wang, and J. Taek Yoon, “Ensemble of data-driven prognostic
algorithms for robust prediction of remaining useful life,” Reliab. Eng. Syst. Saf., vol. 103, pp.
120–135, Jul. 2012, doi: 10.1016/j.ress.2012.03.008.
[21] L. Guo, N. Li, F. Jia, Y. Lei, and J. Lin, “A recurrent neural network based health indicator for
remaining useful life prediction of bearings,” Neurocomputing, vol. 240, pp. 98–109, May
2017, doi: 10.1016/j.neucom.2017.02.045.
[22] T. Han, J. Pang, and A. C. C. Tan, “Remaining useful life prediction of bearing based on
stacked autoencoder and recurrent neural network,” J. Manuf. Syst., vol. 61, pp. 576–591, Oct.
2021, doi: 10.1016/j.jmsy.2021.10.011.
[23] F. Zeng, Y. Li, Y. Jiang, and G. Song, “An online transfer learning-based remaining useful life
prediction method of ball bearings,” Measurement, vol. 176, p. 109201, May 2021, doi:
10.1016/j.measurement.2021.109201.
[24] R. Zemouri, M. Levesque, N. Amyot, C. Hudon, and O. Kokoko, “Deep variational
autoencoder: an efficient tool for PHM frameworks,” 2020 Prognostics and Health
Management Conference (PHM-Besançon), May 2020, pp. 235–240. doi: 10.1109/PHMBesancon49106.2020.00046.
[25] W. Mao, S. Tian, J. Fan, X. Liang, and A. Safian, “Online detection of bearing incipient fault
with semi-supervised architecture and deep feature representation,” J. Manuf. Syst., vol. 55,
pp. 179–198, Apr. 2020, doi: 10.1016/j.jmsy.2020.03.005.
40
[26] W. Mao, D. Zhang, S. Tian, and J. Tang, “Robust detection of bearing early fault based on
deep transfer learning,” Electronics, vol. 9, no. 2, p. 323, Feb. 2020, doi:
10.3390/electronics9020323.
[27] J. Shi, T. Yu, K. Goebel, and D. Wu, “Remaining useful life prediction of bearings using
ensemble learning: The impact of diversity in base learners and features,” J. Comput. Inf. Sci.
Eng., vol. 21, no. 2, Apr. 2021, doi: 10.1115/1.4048215.
[28] M. S. R. Mohd Saufi and K. A. Hassan, “Remaining useful life prediction using an integrated
Laplacian-LSTM network on machinery components,” Appl. Soft Comput., vol. 112, p.
107817, Nov. 2021, doi: 10.1016/j.asoc.2021.107817.
[29] Y. Chen, G. Peng, Z. Zhu, and S. Li, “A novel deep learning method based on attention
mechanism for bearing remaining useful life prediction,” Appl. Soft Comput., vol. 86, p.
105919, Jan. 2020, doi: 10.1016/j.asoc.2019.105919.
[30] J. Zhu, N. Chen, and W. Peng, “Estimation of bearing remaining useful life based on
multiscale convolutional neural network,” IEEE Trans. Ind. Electron., vol. 66, no. 4, pp.
3208–3216, Apr. 2019, doi: 10.1109/TIE.2018.2844856.
[31] Y. Zheng, “Predicting remaining useful life based on Hilbert–Huang entropy with degradation
model,” J. Electr. Comput. Eng., vol. 2019, pp. 1–11, Feb. 2019, doi: 10.1155/2019/3203959.
[32] Y. Yoo and J.-G. Baek, “A novel image feature for the remaining useful lifetime prediction of
bearings based on continuous wavelet transform and convolutional neural network,” Appl.
Sci., vol. 8, no. 7, p. 1102, Jul. 2018, doi: 10.3390/app8071102.
[33] H. Yang, Z. Sun, G. Jiang, F. Zhao, and X. Mei, “Remaining useful life prediction for
machinery by establishing scaled-corrected health indicators,” Measurement, vol. 163, p.
108035, Oct. 2020, doi: 10.1016/j.measurement.2020.108035.
[34] X. Liu, P. Song, C. Yang, C. Hao, and W. Peng, “Prognostics and health management of
bearings based on logarithmic linear recursive least-squares and recursive maximum
likelihood estimation,” IEEE Trans. Ind. Electron., vol. 65, no. 2, pp. 1549–1558, Feb. 2018,
41
doi: 10.1109/TIE.2017.2733469.
[35] J. Yang, S. Yin, Y. Chang, T. Gao, and C. Yang, “An efficient method for monitoring
degradation and predicting the remaining useful life of mechanical rotating components,”
IEEE Trans. Instrum. Meas., vol. 70, pp. 1–14, 2021, doi: 10.1109/TIM.2020.3032218.
[36] L. Ren and W. Lv, “Remaining useful life estimation of rolling bearings based on sparse
representation,” in 2016 7th International Conference on Mechanical and Aerospace
Engineering (ICMAE), Jul. 2016, pp. 209–213. doi: 10.1109/ICMAE.2016.7549536.
[37] G. Zhao, X. Liu, B. Zhang, G. Zhang, G. Niu, and C. Hu, “Bearing health condition prediction
using deep belief network,” Annu. Conf. PHM Soc., vol. 9, 2017, doi:
10.36001/phmconf.2017.v9i1.2484.
[38] A. Savitzky and M. J. E. Golay, “Smoothing and differentiation of data by simplified least
squares procedures.,” Anal. Chem., vol. 36, no. 8, pp. 1627–1639, Jul. 1964, doi:
10.1021/ac60214a047.
[39] D. P. Kingma and M. Welling, “An introduction to variational autoencoders,” Found. Trends
Mach. Learn.,vol. 12, no. 4, pp. 307-392, Jun. 2019, doi: 10.1561/2200000056.
[40] D. C. Montgomery, Statistical quality control: a modern introduction 6/e, 6th ed. Hoboken,
NJ, USA: John Wiley & Sons (Asia), 2008,pp.419–428.
[41] S. Hochreiter and J. Schmidhuber, “Long short-term memory,” Neural Comput., vol. 9, no. 8,
pp. 1735–1780, Nov. 1997, doi: 10.1162/neco.1997.9.8.1735.
[42] M. Schuster and K. K. Paliwal, “Bidirectional recurrent neural networks,” IEEE Trans. Signal
Process., vol. 45, no. 11, pp. 2673–2681, 1997, doi: 10.1109/78.650093.
[43] A. Graves and J. Schmidhuber, “Framewise phoneme classification with bidirectional LSTM
and other neural network architectures,” Neural Networks, vol. 18, no. 5–6, pp. 602–610, Jul.
2005, doi: 10.1016/j.neunet.2005.06.042.
[44] T. Hu, K. Li, H. Ma, H. Sun, and K. Liu, “Quantile forecast of renewable energy generation
based on indicator gradient descent and deep residual BiLSTM,” Control Eng. Pract., vol.
42
114, p. 104863, Sep. 2021, doi: 10.1016/j.conengprac.2021.104863.
[45] F. Wang, Z. E. Amogne, C. Tseng, and J. Chou, “A hybrid method for online cycle life
prediction of lithium‐ion batteries,” Int. J. Energy Res., vol. 46, no. 7, pp. 9080–9096, Jun.
2022, doi: 10.1002/er.7785.
[46] G. Liu and J. Guo, “Bidirectional LSTM with attention mechanism and convolutional layer for
text classification,” Neurocomputing, vol. 337, pp. 325–338, Apr. 2019, doi:
10.1016/j.neucom.2019.01.078.
[47] C. Raffel and D. P. W. Ellis, “Feed-forward networks with attention can solve some long-term
memory Problems,” Int. Conf. Learn. Represent. 2016, Dec. 2016, doi:
10.48550/arXiv.1512.08756v5.
[48] Y. Gal and Z. Ghahramani, “A theoretically grounded application of dropout in recurrent
neural networks,” Adv. neural Inf. Process. Syst. 29, Dec. 2016, doi:
https://doi.org/10.48550/arXiv.1512.05287v5.