本研究探討比較生成對抗網路(Generative Adversarial Network, GAN)和小樣本學習(Few-shot Learning, FSL)在不平衡數據下感應馬達跨負載故障診斷的應用。
首先，利用本研究團研製的感應馬達故障模型，分別為健康、定子匝間短路、轉子斷條、軸承外環損傷和不對心故障，測量於無載、1/4載、半載、3/4載與滿載等不同負載程度下的振動加速度訊號。其次，進行原始數據的前置處理，以選定適合的軸向訊號、圖像轉換之窗口間距與大小以及探討不同的訊號處理方法，可得以莫萊小波(Morlet wavelet)時頻分析並運用卷積神經網路(CNN)模型，於數據充足的情況下，平均故障診斷準確率可達99.4%。再者，探討感應馬達故障診斷中數據不平衡問題，本研究運用具有自注意機制的生成對抗網路(Self-Attention Generative Adversarial Network, SAGAN)能夠學習故障樣本中的局部和全局特徵關聯性，提高生成樣本的真實性。此外，本研究探討於少量數據下，使用基於三元損失(Triplet Loss)孿生神經網路(Siamese Neural Network)小樣本學習，提升診斷表現。
本實驗設計100:1000、50:1000、25:1000、15:1000四種訓練資料不平衡率，實驗結果顯示，運用SAGAN生成額外故障樣本，並以CNN模型進行分類，平均準確率分別為97.6%、94.7%、88.8%和80.9%。相比之下，單純使用原始數據集的基於三元損失孿生神經網路，在不同訓練集資料量下，使用5類別5樣本(5 way 5 shot)測試時，平均準確率分別為98.8%、98.7%、92.3%和65.8%，而在5類別1樣本(5 way 1 shot)進行測試，平均準確率分別為96.9%、95.1%、90.1%、56.8%。可得本研究運用的兩種方法在不平衡率為100:1000、50:1000的感應馬達跨負載故障診斷中皆具有良好的故障診斷準確性，但於不平衡率為15:1000時，則小樣本學習之故障診斷準確率明顯大幅下降。

This thesis compares the application of Generative Adversarial Network and Few-shot Learning for induction motor cross loading fault diagnosis under imbalanced data.
First, the fault model for induction motors is developed by the laboratory, including healthy, stator inter-turn short circuit, rotor broken bar, bearing outer race damage, and misalignment faults. These vibration acceleration signals are measured under different load conditions such as no-load, 1/4 load, half load, 3/4 load, and full load. Next, the raw data is preprocessed to select appropriate signal processing methods, axial signals, and sampling window spacing and size for image transformation. The average fault diagnosis accuracy of using the Morlet wavelet analysis and CNN model in abundant training data is 99.4%. Furthermore, exploring the problem of imbalanced data in motor fault diagnosis. Using Self-Attention Generative Adversarial Network (SAGAN) with self-attention mechanism that learns the local and global feature correlations in samples, thus improving the realism of generated samples. And also using Siamese Neural Network with Triplet Loss for few-shot learning with limited data to enhance the fault diagnostic performance.
In this experiment, four imbalance ratios of 100:1000, 50:1000, 25:1000, and 15:1000 were designed, and the average accuracy of using SAGAN to generate samples and classify them with CNN models was 97.6%, 94.7%, 88.8%, and 80.9%. In contrast, the average accuracy of using Siamese network with the original dataset was 98.8%, 98.7%, 92.3%, and 65.8% for the 5 way 5shot test, and 96.9%, 95.1%, 90.1%, and 56.8% for the 5 way 1 shot test. In the thesis, the two methods have good accuracy in cross loading fault diagnosis of induction motors with imbalance ratios of 50:1000 or 100:1000. However, when the imbalance ratio is 15:1000, the fault diagnosis accuracy of the few-shot learning model significantly decreases.
Keywords: fault diagnosis, imbalanced data, signal processing, generative adversarial networks, few-shot learning

[1] S. Langarica, C. Rüffelmacher and F. Núñez, “An Industrial Internet Application for Real-Time Fault Diagnosis in Industrial Motors,” IEEE Transactions on Automation Science and Engineering, vol. 17, no. 1, pp. 284-295, Jan. 2020.
[2] W. Yu, T. Dillon, F. Mostafa, W. Rahayu and Y. Liu, “A Global Manufacturing Big Data Ecosystem for Fault Detection in Predictive Maintenance,” IEEE Transactions on Industrial Informatics, vol. 16, no. 1, pp. 183-192, Jan. 2020.
[3] Z. Ren, T. Lin, K. Feng, Y. Zhu, Z. Liu and K. Yan, “A Systematic Review on Imbalanced Learning Methods in Intelligent Fault Diagnosis,” IEEE Transactions on Instrumentation and Measurement, vol. 72, pp. 1-35, Art no. 3508535, 2023
[4] J. Goodfellow, J. Pouget-Abadie, M. Mirza, B. Xu, D. Warde-Farley, S. Ozair, A. Courville, and Y. Bengio, “Generative Adversarial Networks,” Advances in Neural Information Processing Systems, 2014, vol. 3, no. 11, 2023.
[5] Y. Wang, Q. Yao, J. -T. Kwok, and L. -M. Ni, “Generalizing from a Few Examples: A Survey on Few-shot Learning,” ACM Computing Survey, vol. 53, no. 3, pp. 1-34, 2020.
[6] L. Ciabattoni, F. Ferracuti, A. Freddi and A. Monteriù, “Statistical Spectral Analysis for Fault Diagnosis of Rotating Machines,” IEEE Transactions on Industrial Electronics, vol. 65, no. 5, pp. 4301-4310, May. 2018.
[7] Y. -G. Lei, J. Lin, Z. -G. He and M. -J. Zuo, “A Review on Empirical Mode Decomposition in Fault Diagnosis of Rotating Machinery,” Mechanical Systems and Signal Processing, vol. 35, no. 1–2, pp. 108-126, Feb. 2013.
[8] T. Yang, H. Pen, Z. Wang and C. S. Chang, “Feature Knowledge Based Fault Detection of Induction Motors Through the Analysis of Stator Current Data,” IEEE Transactions on Instrumentation and Measurement, vol. 65, no. 3, pp. 549-558, 2016.
[9] Z. Huo, Y. Zhang, P. Francq, L. Shu and J. Huang, “Incipient Fault Diagnosis of Roller Bearing Using Optimized Wavelet Transform Based Multi-Speed Vibration Signatures,” IEEE Access, vol. 5, pp. 19442-19456, 2017.
[10] M. Z. Ali, M. N. S. K. Shabbir, X. Liang, Y. Zhang and T. Hu, “Machine Learning-based Fault Diagnosis for Single- and Multi-Faults in Induction Motors Using Measured Stator Currents and Vibration Signals,” IEEE Transactions on Industry Applications, vol. 55, no. 3, pp. 2378-2391, May-June 2019.
[11] Z. -Y Wang, L. -G. Yao and Y. -G. Cai, “Rolling Bearing Fault Diagnosis Using Generalized Refined Composite Multiscale Sample Entropy and Optimized Support Vector Machine,” Measurement, vol 156, Art no. 107574, May. 2020.
[12] N. E. I.Karabadji, I. Khelf, H. Seridi, S. Aridhi, D. Remond and W. Dhifli, “A Data Sampling and Attribute Selection Strategy for Improving Decision Tree Construction,” Expert Systems with Applications, vol. 129, pp. 84-96, Sep. 2019.
[13] S. Tang, S. Yuan and Y. Zhu, “Deep Learning-based Intelligent Fault Diagnosis Methods Toward Rotating Machinery,” IEEE Access, vol. 8, pp. 9335-9346, 2020.
[14] S. Shao, R. Yan, Y. Lu, P. Wang and R. X. Gao, “DCNN-based Multi-Signal Induction Motor Fault Diagnosis,” IEEE Transactions on Instrumentation and Measurement, vol. 69, no. 6, pp. 2658-2669, Jun. 2020.
[15] H. Liang, J. Cao and X. Zhao, “Average Descent Rate Singular Value Decomposition and Two-Dimensional Residual Neural Network for Fault Diagnosis of Rotating Machinery,” IEEE Transactions on Instrumentation and Measurement, vol. 71, pp. 1-16, Art no. 3512616, 2022.
[16] Z. Ren, T. Lin, K. Feng, Y. Zhu, Z. Liu and K. Yan, “A Systematic Review on Imbalanced Learning Methods in Intelligent Fault Diagnosis,” IEEE Transactions on Instrumentation and Measurement, vol. 72, pp. 1-35, Art no. 3508535, 2023.
[17] T. -C. Zhang, J. -L. Chen, F. -D. Li, K. -Y. Zhang, H. -X. Lv, S. -L. He and E. -Y. Xu, “Intelligent Fault Diagnosis of Machines with Small and Imbalanced Data: A State-of-the-Art Review and Possible Extensions,” ISA Transactions, vol. 119, pp. 152-171, Jan. 2022.
[18] X. -Y. Liu, J. Wu and Z. -H. Zhou, “Exploratory Undersampling for Class-Imbalance Learning,” IEEE Transactions on Systems, Man, and Cybernetics, Part B (Cybernetics), vol. 39, no. 2, pp. 539-550, Apr. 2009.
[19] T. Zhu, Y. Lin, and Y. Liu, “Synthetic Minority Oversampling Technique for Multiclass Imbalance Problems,” Pattern Recognition. vol. 72, pp. 327-340, 2017.
[20] G. Douzas, F. Bacao and F. Last, “Improving Imbalanced Learning Through A Heuristic Oversampling Method Based on K-Means and SMOTE,” Information Sciences, vol. 465, pp. 1-20, Oct. 2018.
[21] Z. Wu, W. Lin, B. Fu, J. Guo, Y. Ji and M. Pecht, “A Local Adaptive Minority Selection and Oversampling Method for Class-Imbalanced Fault Diagnostics in Industrial Systems,” IEEE Transactions on Reliability, vol. 69, no. 4, pp. 1195-1206, 2020.
[22] W. Mao, Y. Liu, L. Ding, and Y. Li, “Imbalanced Fault Diagnosis of Rolling Bearing Based on Generative Adversarial Network: A Comparative Study,” IEEE Access, vol. 7, pp. 9515-9530, 2019.
[23] Y. Ding, L. Ma, J. Ma, C. Wang and C. Lu, “A Generative Adversarial Network-based Intelligent Fault Diagnosis Method for Rotating Machinery Under Small Sample Size Conditions,” IEEE Access, vol. 7, pp. 149736-149749, 2019.
[24] Z. Li, T. Zheng, Y. Wang, Z. Cao, Z. Guo and H. Fu, “A Novel Method for Imbalanced Fault Diagnosis of Rotating Machinery Based on Generative Adversarial Networks,” IEEE Transactions on Instrumentation and Measurement, vol. 70, pp. 1-17, Art no. 3500417, 2021.
[25] R. Wang, Z. Chen, S. Zhang and W. Li, “Dual-Attention Generative Adversarial Networks for Fault Diagnosis Under the Class-Imbalanced Conditions, “IEEE Sensors Journal, vol. 22, no. 2, pp. 1474-1485, Jan. 2022.
[26] Luo, J., Huang, J. and Li, H, “A Case Study of Conditional Deep Convolutional Generative Adversarial Networks in Machine Fault Diagnosis,” Journal of Intelligent Manufacturing, vol. 32, pp. 407-425, 2021.
[27] H. Wang, J. Wang, Y. Zhao, Q. Liu, M. Liu and W. Shen, “Few-Shot Learning for Fault Diagnosis with a Dual Graph Neural Network,” IEEE Transactions on Industrial Informatics, vol. 19, no. 2, pp. 1559-1568, Feb. 2023.
[28] Zhang, S. Li, Y. Cui, W. Yang, R. Dong and J. Hu, “Limited Data Rolling Bearing Fault Diagnosis with Few-Shot Learning,” IEEE Access, vol. 7, pp. 110895-110904, 2019.
[29] Z. -J. Ren, Y. -S. Zhu, K. Yan, K. -D. Chen, W. Kang, Y. Yue and D. -W. Gao, “A Novel Model with The Ability of Few-Shot Learning and Quick Updating for Intelligent Fault Diagnosis,” Mechanical Systems and Signal Processing, vol. 138, Art no. 106608, Apr. 2020.
[30] P. Gangsar and R. Tiwari, “Signal Based Condition Monitoring Techniques for Fault Detection and Diagnosis of Induction Motors: A State-of-the-Art Review,” Mechanical Systems and Signal Processing, vol. 144, Art no. 106908, Oct. 2020.
[31] M. E. E. -D. Atta, D. K. Ibrahim and M. I. Gilany, “Broken Bar Fault Detection and Diagnosis Techniques for Induction Motors and Drives: State of the Art,” IEEE Access, vol. 10, pp. 88504-88526, 2022.
[32] H. Henao et al., “Trends in Fault Diagnosis for Electrical Machines: A Review of Diagnostic Techniques,” IEEE Industrial Electronics Magazine, vol. 8, no. 2, pp. 31-42, Jun. 2014.
[33] Siddique, G. S. Yadava and B. Singh, “A Review of Stator Fault Monitoring Techniques of Induction Motors,” IEEE Transactions on Energy Conversion, vol. 20, no. 1, pp. 106-114, Mar. 2005.
[34] G. Mirzaeva and K. I. Saad, “Advanced Diagnosis of Stator Turn-to-Turn Faults and Static Eccentricity in Induction Motors Based on Internal Flux Measurement,” IEEE Transactions on Industry Applications, vol. 54, no. 4, pp. 3961-3970, 2018.
[35] M. Ojaghi, M. Sabouri and J. Faiz, “Performance Analysis of Squirrel-Cage Induction Motors Under Broken Rotor Bar and Stator Inter-Turn Fault Conditions Using Analytical Modeling,” IEEE Transactions on Magnetics, vol. 54, no. 11, pp. 1-5, Nov. 2018, Art no. 8203705.
[36] H. Li, G. Feng, D. Zhen, F. Gu and A. D. Ball, “A Normalized Frequency-Domain Energy Operator for Broken Rotor Bar Fault Diagnosis,” IEEE Transactions on Instrumentation and Measurement, vol. 70, pp. 1-10, Art no. 350, 2021.
[37] Xu, L. Sun and H. Ren, “A New Criterion for the Quantification of Broken Rotor Bars in Induction Motors,” IEEE Transactions on Energy Conversion, vol. 25, no. 1, pp. 100-106, Mar. 2010.
[38] M. Ojaghi, M. Sabouri and J. Faiz, “Analytic Model for Induction Motors Under Localized Bearing Faults,” IEEE Transactions on Energy Conversion, vol. 33, no. 2, pp. 617-626, Jun. 2018.
[39] Y. Shao, R. Kang and J. Liu, “Rolling Bearing Fault Diagnosis Based on the Coherent Demodulation Model,” IEEE Access, vol. 8, pp. 207659-207671, 2020.
[40] J. -L. Lin, Julie Y. -C. Liu, C. -W. Li, L. -F. Tsai and H. -Y. Chung, “Motor Shaft Misalignment Detection Using Multiscale Entropy with Wavelet Denoising,” Expert Systems with Applications, vol. 37, no. 10, pp. 7200-7204, Oct. 2010.
[41] A. Omitaomu, M. K. Jeong, A. B. Badiru and J. W. Hines, “Online Support Vector Regression Approach for the Monitoring of Motor Shaft Misalignment and Feedwater Flow Rate,” IEEE Transactions on Systems, Man, and Cybernetics, Part C (Applications and Reviews), vol. 37, no. 5, pp. 962-970, Sept. 2007.
[42] J. -H. Lee, Y. -H. Lee and N. -S. Kim, “Detection and Analysis of Shaft Misalignment in Application of Production and Logistics Systems Using Motor Current Signature Analysis,” Expert Systems with Applications, vol. 217, Art no. 119463, May. 2023.
[43] PCB Piezotronics, “PCB352C03 Accelerometer, ICP”, https://www.pcb.com/produ
cts?m=352c03.
[44] National Instrument, “PXIe-4492 PXI Sound and Vibration Module”, https://www.
ni.com/zh-tw/support/model.pxie-4492.html.
[45] National Instrument, “SHB4X-8BNC coaxial cable”, https://www.ni.com/zh-tw/su
pport/model.shb4x-8bnc.html.
[46] National Instrument, “NI PXI-8115 Intel Core i5-2510EDual-CorePXI Controller”, https://www.ni.com/zh-tw/support/model.pxie-8115.html.
[47] National Instrument, “PMA-1115 monitor”, https://www.ni.com/zh-tw/support/mo
del.pma-1115.html.
[48] G. Xu, M. Liu, Z. Jiang, W. Shen and C. Huang, “Online Fault Diagnosis Method Based on Transfer Convolutional Neural Networks,” IEEE Transactions on Instrumentation and Measurement, vol. 69, no. 2, pp. 509-520, Feb. 2020.
[49] M.R. Shahriar, T. Ahsan and U. Chong, “Fault Diagnosis of Induction Motors Utilizing Local Binary Pattern-based Texture Analysis,” J Image Video Proc, vol. 2013, pp. 1-11, 2013.
[50] H. -Q. Wang, S. Li, L. -Y. Song, L. -L. Cui, “A Novel Convolutional Neural Network Based Fault Recognition Method via Image Fusion of Multi-vibration-signals,” Computers in Industry, vol. 105, pp. 182-190, 2019.
[51] T. Xie, X. Huang and S. -K. Choi, “Intelligent Mechanical Fault Diagnosis Using Multisensor Fusion and Convolution Neural Network,” IEEE Transactions on Industrial Informatics, vol. 18, no. 5, pp. 3213-3223, May 2022.
[52] 劉人閤 ，「 WGAN-GP與殘差神經網路於有限數據下之感應馬達跨負載故障診斷 」，碩士論文，台灣科技大學電機工程學系， 2022年。
[53] S. K. Gundewar, P. V. Kane, “Bearing Fault Diagnosis Using Time Segmented Fourier Synchrosqueezed Transform Images and Convolution Neural Network,” Measurement, vol. 203, Art no. 111855, 2022.
[54] Y. Xin, S. Li, Z. Zhang, Z. An and J. Wang, “Adaptive Reinforced Empirical Morlet Wavelet Transform and Its Application in Fault Diagnosis of Rotating Machinery,” IEEE Access, vol. 7, pp. 65150-65162, 2019.
[55] H. Zhang, L. Goodfellow, D. Metaxas and A. Odena, “Self-Attention Generative Adversarial Networks,” ArXiv, vol. abs/1805.08318, 2018.
[56] T. Miyato, T. Kataoka, M. Koyama and Y. Yoshida, “Spectral Normalization for Generative Adversarial Networks,” ArXiv, vol. abs/1802.05957, 2018.
[57] M. Heusel, H. Ramsauer, T. Unterthiner, B. Nessler and S. Hochreiter, “GANs Trained by a Two Time-Scale Update Rule Converge to a Local Nash Equilibrium,” ArXiv, vol. abs/1706.08500, 2018.
[58] Kingma and J. Ba, “Adam: A Method for Stochastic Optimization,” ArXiv, vol. abs/1412.6980, 2014.
[59] Gulrajani, F. Ahmed, M. Arjovsky, Vi. Dumoulin and A. Courville, “Improved Training of Wasserstein GANs,” ArXiv, vol. abs/1704.00028, 2017.
[60] Hoffer, N. Ailon, “Deep Metric Learning Using Triplet Network,” ArXiv, vol. abs/1412.6622, 2014.
[61] Schroff, D. Kalenichenko and J. Philbin, “FaceNet: A Unified Embedding for Face Recognition and Clustering,” 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Boston, MA, USA, pp. 815-823, 2015.
[62] Luque, A. Carrasco, A. Martín and A. d. l. Heras, “The Impact of Class Imbalance in Classification Performance Metrics Based on the Binary Confusion Matrix,” Pattern Recognition, vol. 91, pp. 216-231, Jul. 2019.
[63] Kolesnikov, L. Beyer, X. Zhai, J. Puigcerver, J. Yung, S. Gelly and N. Houlsby, “Big Transfer (BiT): General Visual Representation Learning,” ArXiv, vol. abs/1912.11370, 2019.