臺灣博碩士論文加值系統

近年來在工業機台維運中已導入預測性維修(Predictive Maintenance, PdM)之概念，然而與機台協同運作之工業電腦卻沒有導入此觀念。當此工業電腦出現故障時，將會造成莫大經濟損失。本研究透過硬碟中之參數「自我監測、分析及報告技術」(Self-Monitoring Analysis and Reporting Technology, S.M.A.R.T.)以解決硬碟故障所引起之電腦故障。本研究以Elastic Stack建構一可接收S.M.A.R.T.資料之系統，並透過Keras使用長短期記憶模型(Long Short-Term Memory, LSTM)進行硬碟故障預測(Disk Failure Prediction)。本研究成果提出高故障偵測率(Failure Detection Rate, FDR)為0.9632、低假警報率(False Alarm Rate, FAR)為0之模型。

PdM (Predictive Maintenance) has been adopting in production line in recent years. However, PdM has not adopting on Industrial Computer in production line for maintenance tasks. So it will cause production cost increasing as the Industrial Computer is faulted. The hard drive failure is one of main reasons of computer failure. In this study, we perform a hard drive failure prediction model via analyzing S.M.A.R.T. (Self-Monitoring Analysis and Reporting Technology) attributes. By using Elastic Stack architecture to build a system that can collect S.M.A.R.T. information. We introduce a well method based on LSTM (Long Short-Term Memory) networks to estimate hard drive RUL (Remaining Useful Life). Our experiments show the proposed hard drive failure prediction model constructed by real-world datasets could achieve a high FDR (Failure Detection Rate) of 0.9632% with low FAR (False Alarm Rate) of 0%.

摘要 i
Abstract ii
誌謝 iii
目錄 iv
圖目錄 vii
表目錄 viii
第一章 緒論 1
1.1 研究背景 1
1.2 研究動機與目的 3
第二章 文獻探討 4
2.1 S.M.A.R.T.介紹 4
2.1.1 S.M.A.R.T.概述 4
2.1.2 硬碟故障分類 5
2.2 資料中心等級硬碟故障檢測之研究 6
2.3 使用統計、機器學習之硬碟故障檢測 7
2.4 其它硬碟故障檢測文獻探討 9
2.5 硬碟故障預測：計算RUL之研究 11
2.6 ELK實際應用案例 12
第三章 系統架構 & 實驗設計 13
3.1 系統架構介紹 13
3.1.1 系統之主要目的 13
3.1.2 Elastic Stack介紹 13
3.1.3 本研究使用軟體：Logstash、Elasticsearch 14
3.1.4 Elastic Stack相關參數設定 15
3.1.5 Elasticsearch調用方式 15
3.1.6 系統運作流程 15
3.1.7 資料流流程 16
3.2 實驗設計 17
3.2.1 資料集介紹 17
3.2.2 環境介紹 23
3.2.3 資料前處理 (Data Preprocessing) 24
3.3 提出方法 25
3.3.1 健康度分級 25
3.3.2 損失函數 (Loss Function) 27
3.3.3 活化函數 (Activation Function) 28
3.3.4 優化器 (Optimizer) 28
3.3.5 Dropout 28
3.4 模型訓練與測試 29
3.4.1 隱藏層神經元 29
3.4.2 Dropout層 29
3.4.3 優化器 (Optimizer) 29
3.4.4 學習率 (Learning Rate) 30
3.4.5 模型評估方式 (Model Evaluate) 30
第四章 實驗結果與討論 31
4.1 模型測試 31
4.1.1 多對多(30 sequence to 30 sequence)二元分類 31
4.1.2 多對多(30 sequence to 30 sequence)七元分類 32
4.1.3 多對一(30 sequence to 1 output)二元分類-1 33
4.1.4 多對一(30 sequence to 1 output)二元分類-2 34
4.1.5 多對一(30 sequence to 1 output)四元分類 34
4.1.6 多對一(30 sequence to 1 output)六元分類 35
4.1.7 多對一(30 sequence to 1 output)七元分類 36
4.1.8 模型選擇之總結 38
4.2 模型超參數等選擇與測試 39
4.2.1 模型測試-多對一(30 sequence to 1 output)二元分類 39
4.2.2 模型測試-多對一(30 sequence to 1 output)七元分類 42
4.3 實驗結果與討論 45
4.3.1 資料集選擇 45
4.3.2 特徵選取 45
4.3.3 模型測試FDR、FAR之關係 46
第五章 結論與未來工作 47
5.1 結論 47
5.2 未來工作 47
5.2.1 系統層面 47
5.2.2 其他指標 47
參考文獻 48

[1]"自動化系統建置的基本概念-系統維護." https://www.digitimes.com.tw/iot/article.asp?cat=130&cat1=45&cat2=25&id=0000511365_5w92ax8g7hfvfx42agcsh (accessed.
[2]"因硬碟故障而提升之成本." https://www.vertiv.com/en-us/about/news-and-insights/articles/pr-campaigns-reports/benchmark-series/ (accessed.
[3]"產線機台之EAP主機示意圖." https://finance.technews.tw/2018/07/31/kla-tencor-financial-reporting-q4-2018 (accessed.
[4]"Understanding S.M.A.R.T. and S.M.A.R.T. failure and errors." https://support-en.wd.com/app/answers/detail/a_id/12163 (accessed.
[5]J. F. Murray, G. F. Hughes, and K. Kreutz-Delgado, "Machine learning methods for predicting failures in hard drives: A multiple-instance application," Journal of Machine Learning Research, vol. 6, no. May, pp. 783-816, 2005.
[6]"Understanding SMART Technology." http://www.drivehealth.com/smart.html (accessed.
[7]N. Talagala and D. Patterson, An analysis of error behavior in a large storage system. Citeseer, 1999.
[8]E. Pinheiro, W.-D. Weber, and L. A. Barroso, "Failure trends in a large disk drive population," 2007.
[9] L. N. Bairavasundaram, G. R. Goodson, S. Pasupathy, and J. Schindler, "An analysis of latent sector errors in disk drives," in ACM SIGMETRICS Performance Evaluation Review, 2007, vol. 35, no. 1: ACM, pp. 289-300.
[10]V. Agarwal, C. Bhattacharyya, T. Niranjan, and S. Susarla, "Discovering Rules from Disk Events for Predicting Hard Drive Failures," presented at the 2009 International Conference on Machine Learning and Applications, 2009.
[11]V. Agrawal, C. Bhattacharyya, T. Niranjan, and S. Susarla, "Prediction of Hard Drive Failures via Rule Discovery from AutoSupport Data."
[12]S. Ganguly, A. Consul, A. Khan, B. Bussone, J. Richards, and A. Miguel, "A Practical Approach to Hard Disk Failure Prediction in Cloud Platforms: Big Data Model for Failure Management in Datacenters," presented at the 2016 IEEE Second International Conference on Big Data Computing Service and Applications (BigDataService), 2016.
[13] F. Mahdisoltani, I. Stefanovici, and B. Schroeder, "Proactive error prediction to improve storage system reliability," in 2017 {USENIX} Annual Technical Conference ({USENIX}{ATC} 17), 2017, pp. 391-402.
[14] B. Zhu, G. Wang, X. Liu, D. Hu, S. Lin, and J. Ma, "Proactive drive failure prediction for large scale storage systems," in 2013 IEEE 29th Symposium on Mass Storage Systems and Technologies (MSST), 2013: IEEE, pp. 1-5.
[15]W. Yang, D. Hu, Y. Liu, S. Wang, and T. Jiang, "Hard Drive Failure Prediction Using Big Data," presented at the 2015 IEEE 34th Symposium on Reliable Distributed Systems Workshop (SRDSW), 2015.
[16]M. M. Botezatu, I. Giurgiu, J. Bogojeska, and D. Wiesmann, "Predicting Disk Replacement towards Reliable Data Centers," presented at the Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining - KDD '16, 2016.
[17] Z. Tie, H. Zhuang, Q. Luo, and L. Tao, "Statistical method based computer fault prediction model for large computer centers," in 2011 6th International Conference on Computer Science & Education (ICCSE), 2011: IEEE, pp. 960-964.
[18] R. Kumar, S. Vijayakumar, and S. A. Ahamed, "A pragmatic approach to predict hardware failures in storage systems using MPP database and big data technologies," in 2014 IEEE International Advance Computing Conference (IACC), 2014: IEEE, pp. 779-788.
[19]Y. Wang, Q. Miao, E. W. M. Ma, K.-L. Tsui, and M. G. Pecht, "Online Anomaly Detection for Hard Disk Drives Based on Mahalanobis Distance," IEEE Transactions on Reliability, vol. 62, no. 1, pp. 136-145, 2013, doi: 10.1109/tr.2013.2241204.
[20]L. P. Queiroz et al., "A Fault Detection Method for Hard Disk Drives Based on Mixture of Gaussians and Nonparametric Statistics," IEEE Transactions on Industrial Informatics, vol. 13, no. 2, pp. 542-550, 2017, doi: 10.1109/tii.2016.2619180.
[21]J. Li et al., "Hard Drive Failure Prediction Using Classification and Regression Trees," presented at the 2014 44th Annual IEEE/IFIP International Conference on Dependable Systems and Networks, 2014.
[22] C. A. Rincón, J.-F. Pâris, R. Vilalta, A. M. Cheng, and D. D. Long, "Disk failure prediction in heterogeneous environments," in 2017 International Symposium on Performance Evaluation of Computer and Telecommunication Systems (SPECTS), 2017: IEEE, pp. 1-7.
[23]"Hard Drive Data and Stats." https://www.backblaze.com/b2/hard-drive-test-data.html (accessed.
[24] I. C. Chaves, M. R. P. de Paula, L. G. Leite, L. P. Queiroz, J. P. P. Gomes, and J. C. Machado, "Banhfap: A bayesian network based failure prediction approach for hard disk drives," in 2016 5th Brazilian Conference on Intelligent Systems (BRACIS), 2016: IEEE, pp. 427-432.
[25]C. Xu, G. Wang, X. Liu, D. Guo, and T.-Y. Liu, "Health Status Assessment and Failure Prediction for Hard Drives with Recurrent Neural Networks," IEEE Transactions on Computers, vol. 65, no. 11, pp. 3502-3508, 2016, doi: 10.1109/tc.2016.2538237.
[26]F. D. d. S. Lima, G. M. R. Amaral, L. G. d. M. Leite, J. P. P. Gomes, and J. d. C. Machado, "Predicting Failures in Hard Drives with LSTM Networks," presented at the 2017 Brazilian Conference on Intelligent Systems (BRACIS), 2017.
[27] T. Prakash, M. Kakkar, and K. Patel, "Geo-identification of web users through logs using ELK stack," in 2016 6th International Conference-Cloud System and Big Data Engineering (Confluence), 2016: IEEE, pp. 606-610.
[28] I. Y. M. Al-Mahbashi, M. Potdar, and P. Chauhan, "Network security enhancement through effective log analysis using ELK," in 2017 International Conference on Computing Methodologies and Communication (ICCMC), 2017: IEEE, pp. 566-570.
[29] S. J. Son and Y. Kwon, "Performance of ELK stack and commercial system in security log analysis," in 2017 IEEE 13th Malaysia International Conference on Communications (MICC), 2017: IEEE, pp. 187-190.
[30]"Hard Drive SMART Stats." https://www.backblaze.com/blog/hard-drive-smart-stats/ (accessed.