Zobrazit v EDS

Anomaly detection of aerodynamic noise from wind turbine blades based on the CNN–LSTM–AS–UNet model.

Uloženo v:
Podrobná bibliografie
Název: Anomaly detection of aerodynamic noise from wind turbine blades based on the CNN–LSTM–AS–UNet model.
Autoři: Li, Hao, Xi, Yibo, Chen, Hao, Lin, Kuigeng, Wang, Zhenyu
Zdroj: Journal of Civil Structural Health Monitoring; Oct2025, Vol. 15 Issue 7, p3063-3078, 16p
Abstrakt: Wind turbine blades (WTBs) are easily damaged due to prolonged exposure to harsh environments, making anomaly detection crucial for ensuring the safety of wind turbines. To enhance the accuracy of WTB anomaly detection, this study proposes a novel hybrid deep learning model. The proposed model integrates a convolutional neural network (CNN), a long short-term memory model (LSTM), and a segmentation model (U-Net). In this model, the CNN-LSTM is employed to learn and classify data, while U-Net serves as a noise reduction algorithm for image processing. The effectiveness of the hybrid model was validated using the public dataset UrbanSound8K, which comprises 27 h of audio across 10 categories. By training and classifying these 10 audio categories using CNN-LSTM, the model achieves an average accuracy of 96.2%. Subsequently, white noise and sinusoidal noise were superimposed on one of the categories, and the Audio Spectrum U-Net (AS-UNet) was used to denoise the noisy audio. The signal-to-noise ratio (SNR) of the noisy audio was improved by an average of 20 dB. Next, the CNN–LSTM–AS-UNet model was applied to conduct anomaly detection of aerodynamic noise from 20 wind turbines. For aerodynamic audio, which contains multiple noise types, this study employed corresponding noise reduction models for different noises. In addition, an innovative wind turbine-blade energy (WT-BE) standard deviation metric was introduced for anomaly analysis of WTBs. Through comparative analysis of data from 20 wind turbines, it was found that two WTBs were abnormal. These results demonstrate that the proposed model offers significant practical value and holds promise for future applications in WTB anomaly detection. [ABSTRACT FROM AUTHOR]
Copyright of Journal of Civil Structural Health Monitoring is the property of Springer Nature and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)
Databáze: Complementary Index
Popis
Abstrakt:Wind turbine blades (WTBs) are easily damaged due to prolonged exposure to harsh environments, making anomaly detection crucial for ensuring the safety of wind turbines. To enhance the accuracy of WTB anomaly detection, this study proposes a novel hybrid deep learning model. The proposed model integrates a convolutional neural network (CNN), a long short-term memory model (LSTM), and a segmentation model (U-Net). In this model, the CNN-LSTM is employed to learn and classify data, while U-Net serves as a noise reduction algorithm for image processing. The effectiveness of the hybrid model was validated using the public dataset UrbanSound8K, which comprises 27 h of audio across 10 categories. By training and classifying these 10 audio categories using CNN-LSTM, the model achieves an average accuracy of 96.2%. Subsequently, white noise and sinusoidal noise were superimposed on one of the categories, and the Audio Spectrum U-Net (AS-UNet) was used to denoise the noisy audio. The signal-to-noise ratio (SNR) of the noisy audio was improved by an average of 20 dB. Next, the CNN–LSTM–AS-UNet model was applied to conduct anomaly detection of aerodynamic noise from 20 wind turbines. For aerodynamic audio, which contains multiple noise types, this study employed corresponding noise reduction models for different noises. In addition, an innovative wind turbine-blade energy (WT-BE) standard deviation metric was introduced for anomaly analysis of WTBs. Through comparative analysis of data from 20 wind turbines, it was found that two WTBs were abnormal. These results demonstrate that the proposed model offers significant practical value and holds promise for future applications in WTB anomaly detection. [ABSTRACT FROM AUTHOR]
ISSN:21905452
DOI:10.1007/s13349-025-00951-9