基于自注意力和域自适应的风电机组异常状态检测

王晓霞1, 2, 郑肖剑1, 柳璞1, 王荣康1, 王涛3

振动与冲击 ›› 2025, Vol. 44 ›› Issue (10) : 269-277.

PDF(1713 KB)
PDF(1713 KB)
振动与冲击 ›› 2025, Vol. 44 ›› Issue (10) : 269-277.
故障诊断分析

基于自注意力和域自适应的风电机组异常状态检测

  • 王晓霞*1,2,郑肖剑1,柳璞1,王荣康1,王涛3
作者信息 +

Wind turbine abnormal status detection based on self-attention and domain adaptation

  • WANG Xiaoxia*1,2, ZHENG Xiaojian1, LIU Pu1, WANG Rongkang1, WANG Tao3
Author information +
文章历史 +

摘要

针对新建风电机组历史数据不足及不同机组间数据分布差异大的问题,提出一种结合自注意力机制与域自适应网络的风电机组异常状态检测方法。首先,采用编码器-解码器结构对源域和目标域风电机组运行数据进行特征重构,以捕捉潜在的风电模式和领域信息。然后,设计自注意力模块,通过与域判别器的对抗学习提取跨域共享特征,根据跨域信息的匹配度自动加权不同机组的领域信息,实现动态特征重构,从而提升模型对不同机组数据分布变化的适应性。最后,计算重构误差作为异常分数用于异常检测。实际风电机组运行数据的结果表明,该方法在历史数据有限的条件下能够高效地识别风机异常状态,相较于其他深度学习和深度迁移学习方法,显著提升了检测精度。

Abstract

To address the issue of insufficient historical data for newly installed wind turbines and the significant data distribution differences between various turbines, this paper proposes an abnormal status detection method for wind turbines that integrates a self-attention mechanism with domain adaptation networks. Firstly, an encoder-decoder structure is employed to perform feature reconstruction from both source and target domain turbines in order to capture latent wind power patterns and domain-specific information. Then, a self-attention module is designed to extract cross-domain shared features through adversarial learning with a domain discriminator, and domain-specific information is automatically weighted based on the matching degree of cross-domain shared features, enabling dynamic feature reconstruction and thereby improving the model's adaptability to changes in the data distribution across different units. Finally, the reconstruction error is calculated as the abnormal score for anomaly detection. Results from actual wind turbine operation data demonstrate that this method can efficiently identify abnormal data with limited historical data and significantly improve detection accuracy compared to other deep learning and deep transfer learning methods.

关键词

风电机组 / 异常检测 / 域自适应 / 自注意力机制 / 对抗训练

Key words

wind turbines / anomaly detection / domain adaptation / self-attention mechanism / adversarial training

引用本文

导出引用
王晓霞1, 2, 郑肖剑1, 柳璞1, 王荣康1, 王涛3. 基于自注意力和域自适应的风电机组异常状态检测[J]. 振动与冲击, 2025, 44(10): 269-277
WANG Xiaoxia1, 2, ZHENG Xiaojian1, LIU Pu1, WANG Rongkang1, WANG Tao3. Wind turbine abnormal status detection based on self-attention and domain adaptation[J]. Journal of Vibration and Shock, 2025, 44(10): 269-277

参考文献

[1] ANGEL L R M, MATILDE S P. A positive position feedback controller for vibration control of wind turbines[J]. Energy reports, 2023, 9: 1342-1353.
[2] 姚路锦, 王玮, 林宏宇, 等. 计及低碳经济运行的含风电微网群源储协同日前优化调度[J]. 动力工程学报, 2024, 44(03): 462-474.
YAO Lujin, WANG Wei, LIN Hongyu, et al. Optimization scheduling of wind power microgrid group source storage collaboration considering low-carbon economic operation [J] Journal of Power Engineering, 2024, 44 (03): 462-474.
[3] SANTELO T N, DE OLIVEIRA C M R, MACIEL C D, et al. Wind Turbine Failures Review and Trends[J]. Journal of Control, Automation and Electrical Systems, 2021, 33(2): 1-17.
[4] TIAN Y N, ZHONG Y G. Effects of turbine layout spacing and angle on wake interference of floating offshore wind farms[J]. Journal of Mechanical Science and Technology, 2024, 38(3): 1237-1248.
[5] 周凌, 赵前程, 朱岸锋, 等. 基于FISSA-DBN模型的风电机组运行状态监测[J]. 振动测试与诊断, 2023, 43(01): 80-87+199.
ZHOU Ling, ZHAO Qiancheng, ZHU Anfeng, et al. Wind Turbine Operational Status Monitoring Based on FISSA-DBN Model. Vibration Testing and Diagnosis, 2023, 43(01): 80-87+199.
[6] 向玲, 王朋鹤, 李京蓄. 基于CNN-LSTM的风电机组异常状态检测[J]. 振动与冲击, 2021, 40(22): 11-17.
XIANG Ling, WANG Penghe, LI Jingchu. Abnormal state detection of wind turbines based on CNN-LSTM [J] Vibration and Shock, 2021, 40 (22): 11-17.
[7] Meng L, Gao J, Yuan Y, et al. Anomaly detection in wind turbine blades based on PCA and convolutional kernel transform models: employing multivariate SCADA time series analysis[J]. Measurement Science and Technology, 2024, 35(8): 085109.
[8] Li H, Wang Z. Anomaly identification of wind turbine blades based on Mel-Spectrogram Difference feature of aerodynamic noise[J]. Measurement, 2025, 240: 115428.
[9] 康玉祥, 陈果, 盛嘉玖, 等. 低转速航空发动机滚动轴承故障深度异常检测方法[J]. 振动与冲击, 2024, 43(07): 186-195.
Kang Yuxiang, Chen Guo, Sheng Jiajiu, et al. Fault Depth Anomaly Detection Method for Low Speed aero-Engine Rolling Bearing [J]. Journal of Vibration and Shock, 2024, 43(07): 186-195.
[10] 绳晓玲, 韩旭超, 万书亭. 风速分布差异对永磁风力发电机组故障特性的影响分析[J]. 太阳能学报, 2023, 44(02): 123-132.
SHENG Xiaoling, HAN Xuchao, WAN Shuting. Analysis of the influence of wind speed distribution differences on the fault characteristics of permanent magnet wind turbines [J] Journal of Solar Energy, 2023, 44 (02): 123-132.
[11] PAN S J, YANG Q. A Survey on Transfer Learning[J]. IEEE Trans Knowl Data Eng, 2010, 22(10): 1345-1359.
[12] Roelofs C M A, Gück C, Faulstich S. Transfer learning applications for autoencoder-based anomaly detection in wind turbines[J]. Energy and AI, 2024, 17: 100373.
[13] Ming G,Ma S,Xu J, et al. Semi-Supervised Domain Adaptation with Feature Separation for Wind Turbine Anomaly Detection; proceedings of the 2024 IEEE 13th Data Driven Control and Learning Systems Conference (DDCLS), F, 2024 [C]. IEEE.
[14] Altice B, Nazario E, Davis M, et al. Anomaly Detection on Small Wind Turbine Blades Using Deep Learning Algorithms[J]. Energies, 2024, 17(5).
[15] 陈俊夫,皮德常,张强.一种基于迁移学习的遥测数据异常检测方法[J].宇航学报,2021,42(04):522-530.
Chen Junfu, Pidchang, Zhang Qiang.A telemetry data anomaly detection method based on transfer learning [J] Journal of Astronautics, 2021, 42 (04): 522-530.
[16] 纪纲, 刘胜军. SCADA系统在能源供应系统中的应用研究[J]. 自动化仪表, 2024, 45(07): 1-5.
JI Gang, LIU Shengjun. Research on the Application of SCADA System in Energy Supply System [J] Automation Instrumentation, 2024, 45 (07): 1-5.
[17] VASWANI A, SHAZEER N, PARMAR N, et al. Attention is all you need [Z]. Proceedings of the 31st International Conference on Neural Information Processing Systems. Long Beach, California, USA; Curran Associates Inc. 2017: 6000-6010.
[18] 范苍宁, 刘鹏, 肖婷, 等. 深度域适应综述:一般情况与复杂情况[J]. 自动化学报, 2021, 47(03): 515-548.
FAN Cangning, LIU Peng, XIAO Ting, et al. A review of deep domain adaptation: general and complex situations [J] Journal of Automation, 2021, 47 (03): 515-548.
[19] MALHOTRA P,RAMAKRISHNAN A,ANAND G, et al. LSTM-based encoder-decoder for multi-sensor anomaly detection[J]. CoRR, 2016, abs/1607.00148.
[20] ZHANG C X, SONG D J, CHEN Y C, et al. A deep neural network for unsupervised anomaly detection and diagnosis in multivariate time series data[J]. Proceedings of the AAAI conference on artificial intelligence, 2019, 33: 1409-1416.
[21] FAWAZ H I, FORESTIER G, WEBER J, et al. Transfer learning for time series classification[C]// proceedings of the 2018 IEEE international conference on big data (Big Data), WA, USA, 2018.

PDF(1713 KB)

Accesses

Citation

Detail

段落导航
相关文章

/