時間序列多通道卷積神經網路用於軸承剩餘可用壽命預估

以作者查詢圖書館館藏

、以作者查詢臺灣博碩士

、以作者查詢全國書目

、勘誤回報

、線上人數：64

、訪客IP：3.135.241.187

姓名

李睿恩(Juei-En Lee) 查詢紙本館藏

畢業系所

資訊工程學系

論文名稱

時間序列多通道卷積神經網路用於軸承剩餘可用壽命預估
(Time Series Multi-Channel Convolutional Neural Network for Bearing Remaining Useful Life Estimation)

相關論文

★ 以IEEE 802.11為基礎行動隨意無線網路之混合式省電通訊協定	★ 以范諾圖為基礎的對等式網路虛擬環境相鄰節點一致性研究
★ 行動隨意網路可調適及可延展之位置服務協定	★ 同儕式網路虛擬環境高效率互動範圍群播
★ 巨量多人線上遊戲之同儕網路互動範圍語音交談	★ 基於范諾圖之同儕式網路虛擬環境狀態管理
★ 利用多變量分析之多人線上遊戲信任使用者選擇	★ 無位置資訊無線感測網路之覆蓋及連通維持
★ 同儕網路虛擬環境3D串流同儕選擇策略	★ 一個使用802.11與RFID技術的無所不在導覽系統U-Guide之設計與實作
★ 同儕式三維資料串流	★ IM Finder: 透過即時通訊網路線上使用者找尋解答
★ 無位置資訊無線感測網路自走車有向天線導航與協調演算法	★ 多匯點無線感測網路省能及流量分散事件輪廓追蹤
★ 頻寬感知同儕式3D串流	★ 無線感測網路旋轉指向天線定位法

檔案

[Endnote RIS 格式]

[Bibtex 格式]

[相關文章]

[文章引用]

[完整記錄]

[館藏目錄]

[檢視]

[下載]

本電子論文使用權限為同意立即開放。
已達開放權限電子全文僅授權使用者為學術研究之目的，進行個人非營利性質之檢索、閱讀、列印。
請遵守中華民國著作權法之相關規定，切勿任意重製、散佈、改作、轉貼、播送，以免觸法。

摘要(中)

現今全球製造業致力於將工廠藉由工業物聯網、大數據分析、虛實融合系統(Cyber Physical System, CPS)等技術用以實現工業4.0智慧工廠(Smart Factory)，而預診斷及健康管理(Prognostic and Health Management, PHM)是智慧工廠中一項重要的核心系統，透過大數據的蒐集與分析，此系統能讓我們快速的掌握機械的運作情形，提早做出因應的措施。本篇論文著重於發展預診斷及健康管理中的機器剩餘可用壽命(Remaining useful life, RUL)預測技術，利用深度神經網路(Deep Neural Network, DNN)模型預估機械元件的剩餘可用壽命，可避免因元件突然損壞使機器瞬間停止運作而造成重大損失。
本論文提出了時間序列多通道卷積神經網路(Time Series Multi-Channel Convolutional Neural Network, TSMC-CNN)架構對機械設備進行剩餘可用壽命之評估，TSMC-CNN與傳統CNN不同之處在於，傳統CNN主要應用於圖片辨識或影像處理上，而TSMC-CNN將時序性的資料透過多重折疊的疊加處理，讓神經網路能夠提取出長時間序列資料變化的有效特徵，準確的預估機械設備剩餘可用壽命。
本論文以法國研究機構FEMTO-ST在PRONOSTIA實驗平台蒐集的軸承運行資料來驗證我們所提出的TSMC-CNN預測軸承剩餘可用壽命的精確度(accuracy)，且和文獻中所提出的DNN、GBDT、SVM、BP、Gaussian regression、Bayesian Ridge方法做比較，實驗結果顯示，我們提出之TSMC-CNN架構無論是均方根差(RMSE)，或是平均絕對誤差(MAE)結果都是最佳的。

摘要(英)

Today′s global manufacturing industry is committed to transforming traditional factories into industrial 4.0 smart factories through technologies such as Industrial Internet of Things (IIoT), big data analysis, and Cyber Physical System (CPS). The Prognostics and Health Management (PHM) system is one of important systems of the smart factory. Through the collection and analysis of big data, the system can allow users to monitor machinery operation states and health condition in a timely manner so that proper countermeasures can be taken as soon as possible to mitigate potential problems. This study focuses on developing the Remaining Useful Life (RUL) estimation method for the smart factory PHM system. The method can be used to avoid sudden component/machine failures, which may lead to a huge loss.
In this study, we propose a deep learning method using the Time Series Multi-Channel Convolutional Neural Network (TSMC-CNN) architecture for the RUL estimation. Unlike the traditional CNN architecture that is mainly used for image recognition or image processing, the TSMC-CNN architecture divides time-series data into multiple folds and superimpose them altogether to extract relationship between data pieces that are far apart for accurately predicting the RUL of machine/component. The bearing operation data collected by the French research institute FEMTO-ST on the PRONOSTIA experimental platform is used to evaluate the accuracy of the proposed method. The evaluation results are compared with those of the DNN, GBDT, SVM, BP, Gaussian regression, and Bayesian Ridge methods proposed in the literature. The comparisons show that the proposed method is the best in the aspects of both the root mean squared error (RMSE) and the mean absolute error (MAE).

關鍵字(中)

★ 工業4.0
★ 智慧工廠
★ 虛實融合系統
★ 預診斷及健康管理
★ 剩餘可用壽命
★ 卷積神經網路
★ 深度學習
★ 時間序列

關鍵字(英)

★ Industry 4.0
★ smart factory
★ Cyber Physical System
★ prognostics and health management
★ remaining useful life
★ convolutional neural network
★ deep learning
★ time series

論文目次

中文摘要 I
Abstract II
誌謝 III
目錄 IV
圖目錄 VI
表目錄 VIII
一、緒論 1
1.1.研究背景與動機 1
1.2.研究目的與貢獻 2
1.3.論文架構 2
二、背景知識 3
2.1.剩餘可用壽命(Remaining useful life, RUL) 3
2.2.故障預測與健康管理(Prognostic and Health Management) 3
2.3.類神經網路 6
2.3.1.原理 6
2.3.2.架構 7
2.3.3.神經網路學習類別 8
2.3.4.反向傳播演算法(Back-Propagation Algorithm) 9
2.4.深度學習 11
2.4.1.深度學習介紹 11
2.4.2.卷積神經網路(Convolutional Neural Network, CNN) 15
三、問題定義與研究 20
3.1.問題定義 20
3.2.資料集 20
3.3.文獻研究 22
3.3.1.資料集標籤 23
3.3.2.資料特徵萃取 24
3.3.3.深度學習訓練 24
3.3.4.預測評估標準 25
3.3.5.方法比較 25
四、研究方法 27
4.1.標籤定義 27
4.2.時間序列多通道卷積神經網路 27
4.3.網路架構 30
4.3.1.卷積層(Convolutional Layer) 30
4.3.2.池化層(Pooling Layer) 31
4.3.3.訓練最佳化 31
五、實驗與分析 33
5.1.實驗環境 33
5.1.1.硬體設備 33
5.1.2.訓練框架 33
5.2.實驗結果 34
5.3.實驗觀察與分析 39
六、結論與未來展望 40
參考文獻 41

參考文獻

[1] Kothamasu, R., Huang, S. H., & VerDuin, W. H. (2006). System health monitoring and prognostics—a review of current paradigms and practices. The International Journal of Advanced Manufacturing Technology, 28(9-10), 1012-1024.
[2] Shin, J. H., & Jun, H. B. (2015). On condition based maintenance policy. Journal of Computational Design and Engineering, 2(2), 119-127.
[3] Lee, J., Jin, C., Liu, Z., & Ardakani, H. D. (2017). Introduction to data-driven methodologies for prognostics and health management. In Probabilistic prognostics and health management of energy systems (pp. 9-32). Springer, Cham.
[4] Wang, T., Yu, J., Siegel, D., & Lee, J. (2008, October). A similarity-based prognostics approach for remaining useful life estimation of engineered systems. In Prognostics and Health Management, 2008. PHM 2008. International Conference on(pp. 1-6). IEEE.
[5] Nectoux, P., Gouriveau, R., Medjaher, K., Ramasso, E., Chebel-Morello, B., Zerhouni, N., & Varnier, C. (2012, June). PRONOSTIA: An experimental platform for bearings accelerated degradation tests. In IEEE International Conference on Prognostics and Health Management, PHM′12. (pp. 1-8). IEEE Catalog Number: CPF12PHM-CDR.
[6] Ren, L., Cui, J., Sun, Y., & Cheng, X. (2017). Multi-bearing remaining useful life collaborative prediction: A deep learning approach. Journal of Manufacturing Systems, 43, 248-256.
[7] Daigle, M., & Goebel, K. (2010, March). Model-based prognostics under limited sensing. In Aerospace Conference, 2010 IEEE (pp. 1-12). IEEE.
[8] Sun, C., Bisland, S. G., Nguyen, K., & Vu, L. (2006, May). Prognostic/diagnostic health management system (PHM) for fab efficiency. In Advanced Semiconductor Manufacturing Conference, 2006. ASMC 2006. The 17th Annual SEMI/IEEE(pp. 433-438). IEEE.
[9] Grasso, M., Chatterton, S., Pennacchi, P., & Colosimo, B. M. (2016). A data-driven method to enhance vibration signal decomposition for rolling bearing fault analysis. Mechanical Systems and Signal Processing, 81, 126-147.
[10] Si, X. S., Wang, W., Hu, C. H., & Zhou, D. H. (2011). Remaining useful life estimation–a review on the statistical data driven approaches. European journal of operational research, 213(1), 1-14.
[11] LeCun, Y., Touresky, D., Hinton, G., & Sejnowski, T. (1988, June). A theoretical framework for back-propagation. In Proceedings of the 1988 connectionist models summer school(pp. 21-28). CMU, Pittsburgh, Pa: Morgan Kaufmann.
[12] Schmidhuber, J. (2015). Deep learning in neural networks: An overview. Neural networks, 61, 85-117.
[13] The Difference Between Artificial Intelligence, Machine Learning, and Deep Learning:
https://blogs.nvidia.com/blog/2016/07/29/whats-difference-artificial-intelligence-machine-learning-deep-learning-ai/
[14] Szegedy, C., Liu, W., Jia, Y., Sermanet, P., Reed, S., Anguelov, D., ... & Rabinovich, A. (2015, June). Going deeper with convolutions. Cvpr.
[15] He, K., Zhang, X., Ren, S., & Sun, J. (2016). Deep residual learning for image recognition. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 770-778).
[16] Bouvrie, J. (2006). Notes on convolutional neural networks.
[17] LeCun, Y., Bottou, L., Bengio, Y., & Haffner, P. (1998). Gradient-based learning applied to document recognition. Proceedings of the IEEE, 86(11), 2278-2324.
[18] Deep Learning CNN architecture:
https://hackernoon.com/deep-learning-cnns-in-tensorflow-with-gpus-cba6efe0acc2
[19] Liao, L., & Ahn, H. I. (2016). Combining Deep Learning and Survival Analysis for Asset Health Management. International Journal of Prognostics and Health Management.
[20] Babu, G. S., Zhao, P., & Li, X. L. (2016, April). Deep convolutional neural network based regression approach for estimation of remaining useful life. In International conference on database systems for advanced applications (pp. 214-228). Springer, Cham.
[21] Chen, H. (2011). A multiple model prediction algorithm for CNC machine wear PHM. International Journal of Prognostics and Health Management Volume 2 (color), 129.
[22] Medjaher, K., Tobon-Mejia, D. A., & Zerhouni, N. (2012). Remaining useful life estimation of critical components with application to bearings. IEEE Transactions on Reliability, 61(2), 292-302.
[23] IEEE PHM 2012 Prognostic challenge: IEEEPHM2012-Challenge-Details.pdf
[24] Kuo, C. K., Jiang, J. R. (2017). Enhancing Convolutional Neural Network Deep Learning for Remaining Useful Life Estimation.
[25] Friedman, J. H. (2001). Greedy function approximation: a gradient boosting machine. Annals of statistics, 1189-1232.
[26] Di Maio, F., Tsui, K. L., & Zio, E. (2012). Combining relevance vector machines and exponential regression for bearing residual life estimation. Mechanical systems and signal processing, 31, 405-427.
[27] Boškoski, P., Gašperin, M., & Petelin, D. (2012, June). Bearing fault prognostics based on signal complexity and Gaussian process models. In Prognostics and Health Management (PHM), 2012 IEEE Conference on (pp. 1-8). IEEE.
[28] Minka, T. (2000). Bayesian linear regression. Technical report, MIT.
[29] Kingma, D. P., & Ba, J. (2014). Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980.

指導教授

江振瑞(Jehn-Ruey Jiang)

審核日期

2018-6-8

推文