研究生: |
施渝陽 Yu-Yang Shih |
---|---|
論文名稱: |
應用深度卷積生成對抗網路於訓練數據不平衡之感應馬達故障診斷 Fault diagnosis of induction motors with imbalanced data using Deep Convolutional Generative Adversarial Network |
指導教授: |
張宏展
Hong-Chan Chang |
口試委員: |
郭政謙
陳鴻誠 李俊耀 黃維澤 |
學位類別: |
碩士 Master |
系所名稱: |
電資學院 - 電機工程系 Department of Electrical Engineering |
論文出版年: | 2021 |
畢業學年度: | 109 |
語文別: | 中文 |
論文頁數: | 78 |
中文關鍵詞: | 故障診斷 、時頻分析 、生成對抗網路 、數據不平衡 |
外文關鍵詞: | fault diagnosis, time-frequency analysis, generative adversarial network, unbalanced data |
相關次數: | 點閱:311 下載:0 |
分享至: |
查詢本校圖書館目錄 查詢臺灣博碩士論文知識加值系統 勘誤回報 |
在工程應用中,為確保感應馬達運轉時的可靠度和穩定性,使得馬達故障診斷研究成了近年來的熱門話題。然而,現今大多的數據驅動的故障診斷方法皆在充裕且平衡的訓練數據條件下,才有傑出的診斷能力,但在實際應用中擷取的感應馬達訊號,多屬於健康狀態,導致擷取出的訓練集資料多為不平衡的,將大幅降低感應馬達故障診斷模型的性能和實用性。
而本研究由實驗室自製之感應馬達瑕疵模型中,擷取出共五種感應馬達運轉狀態在不同負載下的振動加速度訊號,透過深度卷積生成對抗網路(DCGAN)和卷積神經網路(CNN)兩大學習模型,完成感應馬達故障診斷,旨在解決訓練數據不平衡的問題。在案例分析中分為訓練數據充裕且平衡和訓練數據不足且不平衡兩類進行案例探討,當訓練數據充裕且平衡時,觀察出時頻分析於負載不同時具有故障診斷的優勢,可達95.06%和96.38%;而當訓練數據不足且不平衡時,在任一訊號前處理下,訓練數據越不平衡,所得的測試集準確率皆越低;但透過DCGAN之生成樣本,與實際數據相比具有約80%相似度;且將不平衡的資料集,經由過採樣處理後,能將準確率提高至90%以上,甚至接近實際平衡資料集的準確率,並又以過採樣-平衡(Pro-Balanced)的方法為佳;而在跨負載的測試中,最高仍可得到約85%的診斷結果,說明生成資料學習了具差異性的故障特徵,顯示出DCGAN在學習輸入訊號特徵的能力。
In the engineering applications, in order to ensure the reliability and stability of induction motors during operation, fault diagnosis of induction motors has become a hot topic in recent years. However, data-driven fault diagnosis researches have outstanding diagnostic capabilities only with sufficient training data. And, the captured signals from motor in practical applications are mostly in a healthy state. In view of this, the training set data extracted by the motor is generally unbalanced, which greatly reduces the performance and practicability of the motor fault diagnosis model.
In this study, a defected motor made by the laboratory was used to extract the vibration acceleration signals, and there are five motor operating states under different loads. This paper uses the deep convolutional generative adversarial network (DCGAN) and convolutional neural network (CNN) to complete motor fault diagnosis, aim to solve the problem of unbalanced training data. In the case analysis, it is divided into two categories: training data is sufficient and balanced or insufficient and unbalanced for case study. When the training data is sufficient and balanced, it is observed that time-frequency analysis has the advantage of fault diagnosis when the load is different, up to 95.06 % and 96.38%; and when the training data is insufficient and unbalanced, under any signal pre-processing, the more unbalanced training data is, the lower the accuracy of the classification. However, the samples generated by DCGAN have about 80% similarity of the actual data. After the over-sampling processing, the unbalanced data set can get 90% diagnostic ability, which is close to the accuracy of the actual balanced data set. And the Pro-Balanced method is much better. In the test of cross-load, it obtained about 85% of the diagnosis results, indicating that the generated data from DCGAN has learned the different fault characteristics. After all, it showed the ability of DCGAN to learn the characteristics of the input signal.
參考文獻
[1] S. Orey and W. E. Pruitt, “Sample Functions of the N-Parameter Wiener Process”, Ann. Probab., vol. 1, no. 1, pp. 138-163, Feb. 1973.
[2] F. Gustafsson, F. Gunnarsson, N. Bergman, U. Forssell, J. Jansson, R. Karlsson, and P.-J. Nordlund, “Particle filters for positioning, navigation, and tracking”, IEEE Trans. Signal Process., vol. 50, no. 2, pp.425-437, Feb. 2002.
[3] I. Arasaratnam and S. Haykin, “Cubature Kalman Filters”, IEEE Trans. Autom. Control, nol. 54, no.6, Jun. 2009.
[4] W. Zhang, C. Li, G. Peng, Y. Chen, and Z. Zhang, "A deep convolutional neural network with new training methods for bearing fault diagnosis under noisy environment and different working load", Mech. Syst. Signal Process., vol. 100, pp. 439-453, 2018.
[5] X. Yuan, L. Li, and Y. Wang, "Nonlinear dynamic soft sensor modeling with supervised long short-term memory network", IEEE Trans. Ind. Informat., vol. 16, no.5 May 2020.
[6] N. V. Chawla, L. O. Hall, K. W. Bowyer, and W. P. Kegelmeyer, "SMOTE: Synthetic Minority Oversampling Technique," J. Artificial Intelligence Research, vol. 16, pp. 321-357, 2002.
[7] H. He, Y. Bai, E.A. Garcia, and S. Li, "ADASYN: Adaptive Synthetic Sampling Approach for Imbalanced Learning", Proc. Int'l Joint Conf. Neural Networks, pp. 1322-1328, 2008.
[8] I. Goodfellow et al., "Generative adversarial nets", Proc. Adv. Neural Inf. Process. Syst., pp. 2672-2680, 2014.
[9] W. Zhang, X. Li, X.-D. Jia, H. Ma, Z. Luo, and X. Li, "Machinery fault diagnosis with imbalanced data using deep generative adversarial networks", Measurement, vol. 152, Feb. 2020.
[10] Q. Guo, Y. Li, Y. Song, D. Wang, and W. Chen, "Intelligent fault diagnosis method based on full 1-D convolutional generative adversarial network", IEEE Trans. Ind. Informat., vol. 16, no. 3, pp. 2044-2053, Mar. 2020.
[11] Z. Wang, J. Wang, and Y. Wang, "An intelligent diagnosis scheme based on generative adversarial learning deep neural networks and its application to planetary gearbox fault pattern recognition", Neurocomputing, vol. 310, pp. 213-222, Oct. 2018.
[12] H. Tao, P. Wang, Y. Chen, V. Stojanovic, and H. Yang, "An unsupervised fault diagnosis method for rolling bearing using STFT and generative neural networks", J. Franklin Inst., vol.357, no.11, pp. 7289-7307, Jul. 2020.
[13] P.F. Liang, C. Deng, J. Wu, Z.X. Yang, J.X. Zhu, and Z.H. Zhang, "Single and simultaneous fault diagnosis of gearbox via a semi-supervised and high-accuracy adversarial learning framework", Knowl.-Based Syst., vol.198, Jun. 2020.
[14] Pan T.Y., Chen J.L., Xie J.S., Chang Y.H., and Zhou Z.T., " Intelligent fault identification for industrial automation system via multi-scale convolutional generative adversarial network with partially labeled samples", ISA Trans., vol.101, pp. 379-389, Jun. 2020.
[15] Q. Liu, G. Ma, and C. Cheng, " Data fusion generative adversarial network for multi-class imbalanced fault diagnosis of rotating machinery", IEEE Access, vol.8, pp.70111-70124, April 2020
[16] .X. Li, W. Zhang, and Q. Ding, "Cross-domain fault diagnosis of rolling element bearings using deep generative neural networks", IEEE Trans. Ind. Electron., vol. 66, no. 7, pp. 5525-5534, Jul. 2019.
[17] G. K. Singh and S. A. S. Al Kazzaz, "Induction machine drive condition monitoring and diagnostic research—A survey", Elect. Power Syst. Res., vol. 64, no. 2, pp. 145-158, Feb. 2003.
[18] W. S. Gongora, H. V. D. Silva, A. Goedtel, W. F. Godoy, and S. A. Q. da Silva, "Neural approach for bearing fault detection in three phase induction motors", Proc. 9th IEEE Int. Symp. Diagnostics Electr. Mach. Power Electron. Drives (SDEMPED), pp. 566-572, Aug. 2013.
[19] Y. Lei, B. Yang, X. Jiang, F. Jia, N. Li, and A. K. Nandi, "Applications of machine learning to machine fault diagnosis: A review and roadmap", Mech. Syst. Signal Process., vol. 138, Apr. 2020.
[20] L. Wen, X. Li, L. Gao, and Y. Zhang, "A new convolutional neural network based data-driven fault diagnosis method", IEEE Trans. Ind. Electron., vol. 65, no. 7, pp. 5990-5998, Jul. 2018.
[21] S.-H. Liao, "Expert system methodologies and applications: A decade review from 1995 to 2004", Expert Syst. With Applic., vol. 28, no. 1, pp. 93-103, 2005.
[22] A. Widodo and B. Yang, "Support vector machine in machine condition monitoring and fault diagnosis", Mech. Syst. Signal Process., vol. 21, no. 6, pp. 2560-2574, Aug. 2007.
[23] R. Liu, B. Yang, E. Zio, and X. Chen, "Artificial intelligence for fault diagnosis of rotating machinery: A review", Mech. Syst. Signal Process., vol. 108, pp. 33-47, Aug. 2018.
[24] L. Jiang, Z. Ge, and Z. Song, "Semi-supervised fault classification based on dynamic Sparse Stacked auto-encoders model", Chemometr. Intell. Lab. Syst., vol. 168, pp. 72-83, 2017.
[25] S. Dong et al., "Design and application of unsupervised convolutional neural networks integrated with deep belief networks for mechanical fault diagnosis", Proc. Prognostics Syst. Health Manage. Conf. (PHM-Harbin), pp. 1-7, 2017.
[26] K. He, X. Zhang, S. Ren, and J. Sun, "Deep residual learning for image recognition", 2015.
[27] B. Yang, Y. Lei, F. Jia, and S. Xing, "An intelligent fault diagnosis approach based on transfer learning from laboratory bearings to locomotive bearings", Mech. Syst. Signal Process., vol. 122, pp. 692-706, 2019.
[28] L. Guo, Y. Lei, S. Xing, T. Yan, and N. Li, "Deep convolutional transfer learning network: A new method for intelligent fault diagnosis of machines with unlabeled data", IEEE Trans. Ind. Electron., vol. 66, no. 9, pp. 7316-7325, Sep. 2019.
[29] F. Shen, C. Chen, R.-Q. Yan, and R. X. Gao, "Bearing fault diagnosis based on SVD feature extraction and transfer learning classification", Proc. Prognostics Syst. Health Manage. Conf., pp. 1-6, 2015.
[30] M. J. Hasan and J.-M. Kim, "Bearing fault diagnosis under variable rotational speeds using stockwell transform-based vibration imaging and transfer learning", Appl. Sci., vol. 8, no. 12, pp. 2357, Nov. 2018.
[31] R. Zhao, R. Yan, Z. Chen, K. Mao, P. Wang, and R. X. Gao, "Deep learning and its applications to machine health monitoring", Mech. Syst. Signal Process., vol. 115, pp. 213-237, Jan. 2019.
[32] 林芝以,「運用電氣訊號於感應馬達狀態監測與故障診斷之研究」,碩士論文,台灣科技大學電機工程學系,2016年。
[33] G. Lemaître, F. Nogueira, and C. K. Aridas, "Imbalanced-learn: A python toolbox to tackle the curse of imbalanced datasets in machine learning", J. Mach. Learn. Res., vol. 18, no. 17, pp. 1-5, 2017.
[34] A. Radford, L. Metz and S. Chintala, Unsupervised representation learning with deep convolutional generative adversarial networks, 2015, [online] Available: https://arxiv.org/abs/1511.06434.
[35] 曾思憲,「感應馬達主動型狀態估測系統之研發」,碩士論文,台灣科技大學電機工程學系,2018年。