针对风力机行星齿轮箱异常状态识别任务中难以获取带标签数据以训练分类模型的问题,提出一种无监督的自动化检测方法。首先提取原始振动信号的对数梅尔能量特征作为训练数据,将其输入到基于U-net自动编码器为核心的无监督异常状态识别模型中进行训练,进而基于模型输入与输出间的重构误差设定齿轮箱健康状态的判定临界值,从而完成异常状态的识别任务。同时使用出厂齿轮箱测试数据与山西羊头崖某风场风力机实际运行数据验证提出模型的可靠性。对于出厂齿轮箱,同时使用基于频谱振幅调制的信号处理方法进行双重验证。验证结果表明,所提方法在工厂数据测试集和风场数据测试集上的识别准确率达到93.34%,验证所提无监督方法能够自动化地正确分离异常状态风力机齿轮箱。
Abstract
Significant progress has been made in the fault diagnosis of wind turbine planetary gearboxes based on deep learning. To address the difficulty to obtain labelled data for training classification models in the task of identifying abnormal states of planetary gearboxes, this paper proposes an unsupervised automatic detection method, in which the Log Mel-band Energies features of the original vibration signals are firstly extracted and input as training data into an unsupervised abnormal state recognition model based on U-net autoencoder as the core for training. Furthermore, based on the reconstruction error between the model inputs and the outputs, a judgment threshold for the health state of the gearbox is set to complete the abnormal state identification task. In this paper, both factory test data of gearboxes and actual operation data of wind turbines from a wind farm in Yangtouya, Shanxi Province are used to validate the reliability of the proposed model. For factory gearboxes, a signal processing method based on spectral amplitude modulation is simultaneously adopted for dual verification. The proposed method achieved an identification accuracy of 93.34% on the factory dataset and the wind farm dataset. This proves that the proposed unsupervised method can automatically and correctly separate abnormal state wind turbine gearboxes.
关键词
无监督学习 /
风电机组 /
齿轮箱 /
异常检测 /
对数梅尔能量 /
自编码器 /
频谱振幅调制
Key words
unsupervised learning /
wind turbines /
gearboxes /
anomaly detection /
Log Mel-band energies /
autoencoder /
spectral amplitude modulation
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1] 刘少康, 武英杰, 田野, 等. 基于振动信号耦合调制新模型和参数辨识的风电齿轮箱齿圈故障监测[J]. 仪器仪表学报, 2022, 43(10): 260-269.
Liu S K, Wu Y J, Tian Y, et al.Fault monitoring of ring gear of wind turbine gearbox based on coupling modulation new model of vibration signal and parameter identification[J]. Chinese Journal of Scientific Instrument, 2022, 43(10): 260-269.
[2] Yu X N, Feng Z P, Liang M.Analytical vibration signal model and signature analysis in resonance region for planetary gearbox fault diagnosis[J]. Journal of Sound and Vibration, 2021, 498: 115962.
[3] Feng Z P, Gao A R, Li K Q, et al.Planetary gearbox fault diagnosis via rotary encoder signal analysis[J]. Mechanical Systems and Signal Processing, 2021, 149: 107325.
[4] Zhao D Z, Cui L L.Scaling operator demodulation spectrum-based planetary gearbox fault diagnosis method under variable speed conditions[J]. Structural Health Monitoring, 2023, 22(4): 2579-2596.
[5] Li W H, Huang R Y, Li J P, et al.A perspective survey on deep transfer learning for fault diagnosis in industrial scenarios: theories, applications and challenges[J]. Mechanical Systems and Signal Processing, 2022, 167: 108487.
[6] Gao Z W, Cecati C, Ding S X.A survey of fault diagnosis and fault-tolerant techniques: part Ⅰ: fault diagnosis with model-based and signal-based approaches[J]. IEEE Transactions on Industrial Electronics, 2015, 62(6): 3757-3767.
[7] Xia J Y, Huang R Y, Liao Y X, et al.Digital twin-assisted gearbox dynamic model updating toward fault diagnosis[J]. Frontiers of Mechanical Engineering, 2023, 18(2): 32.
[8] Booyse W, Wilke D N, Heyns S.Deep digital twins for detection, diagnostics and prognostics[J]. Mechanical Systems and Signal Processing, 2020, 140: 106612.
[9] Wang Y, Sun W L, Liu L Q, et al.Fault diagnosis of wind turbine planetary gear based on a digital twin[J]. Applied Sciences, 2023, 13(8): 4776.
[10] Baldi P.Autoencoders, unsupervised learning, and deep architectures[C]//Proceedings of ICML workshop on unsupervised and transfer learning. JMLR Workshop and Conference Proceedings, 2012: 37-49.
[11] 刘家瑞, 杨国田, 杨锡运. 基于深度卷积自编码器的风电机组故障预警方法研究[J]. 太阳能学报, 2022, 43(11): 215-223.
Liu J R, Yang G T, Yang X Y.Research on wind turbine fault warning method based on deep convolution auto-encoder[J]. Acta Energiae Solaris Sinica, 2022, 43(11): 215-223.
[12] Yang L X, Zhang Z J.A conditional convolutional autoencoder-based method for monitoring wind turbine blade breakages[J]. IEEE Transactions on Industrial Informatics, 2021, 17(9): 6390-6398.
[13] Sun Z X, Sun H X.Stacked denoising autoencoder with density-grid based clustering method for detecting outlier of wind turbine components[J]. IEEE Access, 2019, 7: 13078-13091.
[14] Wu P, Wang Y X, Zhang X J, et al.Wind turbine blade breakage monitoring with mogrifier LSTM autoencoder[J]. IEEE Transactions on Instrumentation and Measurement, 2023, 72: 3534610.
[15] Moshrefzadeh A, Fasana A, Antoni J.The spectral amplitude modulation: a nonlinear filtering process for diagnosis of rolling element bearings[J]. Mechanical Systems and Signal Processing, 2019, 132: 253-276.
[16] Sushil M, Šuster S, Luyckx K, et al.Patient representation learning and interpretable evaluation using clinical notes[J]. Journal of Biomedical Informatics, 2018, 84: 103-113.
基金
国家自然科学基金(52475548); 中国博士后科学基金（2023M740429）; 重庆市教委科学技术研究项目（KJQN202400753）; 重庆市研究生科研创新项目（CYS260569）
PDF(3258 KB)
