End-to-End Deep Learning Approach for Wind Turbine Bearing Fault Detection From Acoustic Data
Abstract
Wind energy generation through wind turbines has become increasingly attractive as a clean and renewable energy source. Effective maintenance of wind turbines is crucial, as failure can result in significant economic losses and damage to equipment due to unplanned downtime. Nevertheless, ensuring effective maintenance remains challenging because these systems usually operate in severe and remote environments. Considering that rolling bearings are among the most necessary mechanical components in wind turbines, accurately detecting faults in these bearings is important for ensuring the regular and reliable operation of the equipment. This paper proposes a deep learning-based monitoring architecture that utilizes acoustic signals emitted from rolling bearings to detect faults in wind turbines. Using real-world data collected via microphones and a Raspberry Pi system, we constructed a structured and manually annotated dataset. A convolutional neural network (CNN) model was trained on mel-spectrogram representations to distinguish between healthy and faulty operational states. The system achieved promising performance, with 83.62% accuracy, 87.40% F1-score, 95% AUC-ROC, and 98% precision-recall AUC. The proposed end-to-end pipeline integrates data acquisition, pre-processing, classification, and confidence-based decision thresholds, making it suitable for deployment in operational monitoring scenarios. These results demonstrate the viability of audio-based fault detection as a scalable and non-invasive solution for predictive maintenance in wind energy systems.
Keywords:
Audible noise, Fault detection, Artificial Intelligence, Wind Turbine
References
Arohan Ajit, Koustav Acharya, and Abhishek Samanta. 2020. A review of convolutional neural networks. In 2020 International Conference on Emerging Trends In Information Technology And Engineering (ic-ETITE). IEEE, 1–5.
Aiman Al-Sabaawi, HassanMIbrahim, Zinah Mohsin Arkah, Muthana Al-Amidie, and Laith Alzubaidi. 2020. Amended convolutional neural network with global average pooling for image classification. In International Conference on Intelligent Systems Design and Applications. Springer, 171–180.
Pierre Baldi and Peter J Sadowski. 2013. Understanding dropout. Advances in neural information processing systems 26 (2013).
Davide Chicco and Giuseppe Jurman. 2020. The advantages of the Matthews correlation coefficient (MCC) over F1 score and accuracy in binary classification evaluation. BMC genomics 21, 1 (2020), 6.
Yan Martins B Gurevitz Cunha, José Matheus C Boaro, Daniel de Sousa Moraes, Pedro Cutrim dos Santos, Polyana Bezerra da Costa, Antonio José Grandson Busson, Julio Cesar Duarte, and Sérgio Colcher. 2024. An Ensemble Approach to Facial Deepfake Detection Using Self-Supervised Features. In Brazilian Symposium on Multimedia and the Web (WebMedia). SBC, 37–44.
Shaveta Dargan, Munish Kumar, Maruthi Rohit Ayyagari, and Gulshan Kumar. 2020. A survey of deep learning and its applications: a new paradigm to machine learning. Archives of computational methods in engineering 27, 4 (2020), 1071–1092.
Rajvikram Madurai Elavarasan, Rishi Pugazhendhi, Muhammad Irfan, Lucian Mihet-Popa, Irfan Ahmad Khan, and Pietro Elia Campana. 2022. State-of-the-art sustainable approaches for deeper decarbonization in Europe–An endowment to climate neutral vision. Renewable and Sustainable Energy Reviews 159 (2022), 112204.
Tom Fawcett. 2006. An introduction to ROC analysis. Pattern recognition letters 27, 8 (2006), 861–874.
Mathaus Ferreira da Silva, Juliano Emir Nunes Masson, Murillo Ferreira dos Santos, William Rodrigues Silva, Iuri Wladimir Molina, and Gabriel Miguel Castro Martins. 2025. Audible Noise Evaluation in Wind Turbines Through Artificial Intelligence Techniques. Sensors 25, 5 (2025), 1492.
Peter Flach. 2012. Machine learning: the art and science of algorithms that make sense of data. Cambridge university press.
Zafar Hameed, Young-Sun Hong, YM Cho, SH Ahn, and Chul Ki Song. 2009. Condition monitoring and fault detection of wind turbines and related algorithms: A review. Renewable and Sustainable energy reviews 13, 1 (2009), 1–39.
Haibo He and Edwardo A Garcia. 2009. Learning from imbalanced data. IEEE Transactions on knowledge and data engineering 21, 9 (2009), 1263–1284.
Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. 2015. Delving deep into rectifiers: Surpassing human-level performance on imagenet classification. In Proceedings of the IEEE International Conference on Computer Vision. 1026–1034.
Jürgen Herp, Mohammad H Ramezani, Martin Bach-Andersen, Niels L Pedersen, and Esmaeil S Nadimi. 2018. Bayesian state prediction of wind turbine bearing failure. Renewable Energy 116 (2018), 164–172.
Maximilian Hoh, Alfred Schöttl, Henry Schaub, and Franz Wenninger. 2022. A generative model for anomaly detection in time series data. Procedia Computer Science 200 (2022), 629–637.
Sergey Ioffe and Christian Szegedy. 2015. Batch normalization: Accelerating deep network training by reducing internal covariate shift. In International Conference on Machine Learning. pmlr, 448–456.
László A Jeni, Jeffrey F Cohn, and Fernando De La Torre. 2013. Facing imbalanced data–recommendations for the use of performance metrics. In 2013 Humaine association conference on affective computing and intelligent interaction. IEEE, 245–251.
Baljinder Kaur and Jaskirat Singh. 2021. Audio classification: Environmental sounds classification. (2021). [link]
Asifullah Khan, Anabia Sohail, Umme Zahoora, and Aqsa Saeed Qureshi. 2020. A survey of the recent architectures of deep convolutional neural networks. Artificial intelligence review 53, 8 (2020), 5455–5516.
Yann LeCun, Yoshua Bengio, and Geoffrey Hinton. 2015. Deep learning. nature 521, 7553 (2015), 436–444.
Yann LeCun, Bernhard Boser, John Denker, Donnie Henderson, Richard Howard, Wayne Hubbard, and Lawrence Jackel. 1989. Handwritten digit recognition with a back-propagation network. Advances in neural information processing systems 2 (1989).
Li Li, Miloš Doroslovački, and Murray H Loew. 2020. Approximating the gradient of cross-entropy loss function. IEEE access 8 (2020), 111626–111635.
Xiaofan Lin, Cong Zhao, and Wei Pan. 2017. Towards accurate binary convolutional neural network. Advances in neural information processing systems 30 (2017).
Majid Memari, Praveen Shakya, Mohammad Shekaramiz, Abdennour C Seibi, and Mohammad AS Masoum. 2024. Review on the advancements in wind turbine blade inspection: Integrating drone and deep learning technologies for enhanced defect detection. IEEE Access (2024).
Vinod Nair and Geoffrey E Hinton. 2010. Rectified linear units improve restricted boltzmann machines. In Proceedings of the 27th International Conference on Machine Learning (ICML-10). 807–814.
Vladimir Nasteski. 2017. An overview of the supervised machine learning methods. Horizons. b 4, 51-62 (2017), 56.
Otacílio de A Ramos Neto, Rafael C Chaves, Alysson P Nascimento, and Ruan D Gomes. 2024. Middleware para aplicações distribuídas de vídeo com suporte à computação na borda na Indústria 4.0. In Brazilian Symposium on Multimedia and the Web (WebMedia). SBC, 215–222.
Han Peng, Hai Zhang, Yisa Fan, Linjian Shangguan, and Yang Yang. 2022. A review of research on wind turbine bearings’ failure analysis and fault diagnosis. Lubricants 11, 1 (2022), 14.
David MW Powers. 2020. Evaluation: from precision, recall and F-measure to ROC, informedness, markedness and correlation. arXiv preprint arXiv:2010.16061 (2020).
Wei Qiao and Dingguo Lu. 2015. A survey on wind turbine condition monitoring and fault diagnosis—Part II: Signals and signal processing methods. IEEE Transactions on Industrial Electronics 62, 10 (2015), 6546–6557.
Christine Röckmann, Sander Lagerveld, and John Stavenuiter. 2017. Operation and maintenance costs of offshore wind farms and potential multi-use platforms in the Dutch North Sea. In Aquaculture Perspective of Multi-Use Sites in the Open Ocean: The Untapped Potential for Marine Resources in the Anthropocene. Springer International Publishing Cham, 97–113.
Takaya Saito and Marc Rehmsmeier. 2015. The precision-recall plot is more informative than the ROC plot when evaluating binary classifiers on imbalanced datasets. PloS one 10, 3 (2015), e0118432.
Marina Sokolova and Guy Lapalme. 2009. A systematic analysis of performance measures for classification tasks. Information processing & management 45, 4 (2009), 427–437.
Bouno Toshio, Yuji Toshifumi, Hamada Tsugio, and Hideaki Toya. 2005. Failure forecast diagnosis of small wind turbine using acoustic emission sensor. KIEE International Transaction on Electrical Machinery and Energy Conversion Systems 5, 1 (2005), 78–83.
Xin Wang, Dongxing Mao, and Xiaodong Li. 2021. Bearing fault diagnosis based on vibro-acoustic data fusion and 1D-CNN network. Measurement 173 (2021), 108518.
Khalid Zaman, Melike Sah, Cem Direkoglu, and Masashi Unoki. 2023. A survey of audio classification using deep learning. IEEE Access 11 (2023), 106620–106649.
Pinjia Zhang and Delong Lu. 2019. A survey of condition monitoring and fault diagnosis toward integrated O&M for wind turbines. Energies 12, 14 (2019), 2801.
Yang Zhang, Maciej Radzieński, Sidney Xue, Ziping Wang, and Wiesław Ostachowicz. 2024. A method for detecting cracks on the trailing edge of wind turbine blades based on aeroacoustic noise analysis. Structural Health Monitoring (2024), 14759217241303452.
Aiman Al-Sabaawi, HassanMIbrahim, Zinah Mohsin Arkah, Muthana Al-Amidie, and Laith Alzubaidi. 2020. Amended convolutional neural network with global average pooling for image classification. In International Conference on Intelligent Systems Design and Applications. Springer, 171–180.
Pierre Baldi and Peter J Sadowski. 2013. Understanding dropout. Advances in neural information processing systems 26 (2013).
Davide Chicco and Giuseppe Jurman. 2020. The advantages of the Matthews correlation coefficient (MCC) over F1 score and accuracy in binary classification evaluation. BMC genomics 21, 1 (2020), 6.
Yan Martins B Gurevitz Cunha, José Matheus C Boaro, Daniel de Sousa Moraes, Pedro Cutrim dos Santos, Polyana Bezerra da Costa, Antonio José Grandson Busson, Julio Cesar Duarte, and Sérgio Colcher. 2024. An Ensemble Approach to Facial Deepfake Detection Using Self-Supervised Features. In Brazilian Symposium on Multimedia and the Web (WebMedia). SBC, 37–44.
Shaveta Dargan, Munish Kumar, Maruthi Rohit Ayyagari, and Gulshan Kumar. 2020. A survey of deep learning and its applications: a new paradigm to machine learning. Archives of computational methods in engineering 27, 4 (2020), 1071–1092.
Rajvikram Madurai Elavarasan, Rishi Pugazhendhi, Muhammad Irfan, Lucian Mihet-Popa, Irfan Ahmad Khan, and Pietro Elia Campana. 2022. State-of-the-art sustainable approaches for deeper decarbonization in Europe–An endowment to climate neutral vision. Renewable and Sustainable Energy Reviews 159 (2022), 112204.
Tom Fawcett. 2006. An introduction to ROC analysis. Pattern recognition letters 27, 8 (2006), 861–874.
Mathaus Ferreira da Silva, Juliano Emir Nunes Masson, Murillo Ferreira dos Santos, William Rodrigues Silva, Iuri Wladimir Molina, and Gabriel Miguel Castro Martins. 2025. Audible Noise Evaluation in Wind Turbines Through Artificial Intelligence Techniques. Sensors 25, 5 (2025), 1492.
Peter Flach. 2012. Machine learning: the art and science of algorithms that make sense of data. Cambridge university press.
Zafar Hameed, Young-Sun Hong, YM Cho, SH Ahn, and Chul Ki Song. 2009. Condition monitoring and fault detection of wind turbines and related algorithms: A review. Renewable and Sustainable energy reviews 13, 1 (2009), 1–39.
Haibo He and Edwardo A Garcia. 2009. Learning from imbalanced data. IEEE Transactions on knowledge and data engineering 21, 9 (2009), 1263–1284.
Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. 2015. Delving deep into rectifiers: Surpassing human-level performance on imagenet classification. In Proceedings of the IEEE International Conference on Computer Vision. 1026–1034.
Jürgen Herp, Mohammad H Ramezani, Martin Bach-Andersen, Niels L Pedersen, and Esmaeil S Nadimi. 2018. Bayesian state prediction of wind turbine bearing failure. Renewable Energy 116 (2018), 164–172.
Maximilian Hoh, Alfred Schöttl, Henry Schaub, and Franz Wenninger. 2022. A generative model for anomaly detection in time series data. Procedia Computer Science 200 (2022), 629–637.
Sergey Ioffe and Christian Szegedy. 2015. Batch normalization: Accelerating deep network training by reducing internal covariate shift. In International Conference on Machine Learning. pmlr, 448–456.
László A Jeni, Jeffrey F Cohn, and Fernando De La Torre. 2013. Facing imbalanced data–recommendations for the use of performance metrics. In 2013 Humaine association conference on affective computing and intelligent interaction. IEEE, 245–251.
Baljinder Kaur and Jaskirat Singh. 2021. Audio classification: Environmental sounds classification. (2021). [link]
Asifullah Khan, Anabia Sohail, Umme Zahoora, and Aqsa Saeed Qureshi. 2020. A survey of the recent architectures of deep convolutional neural networks. Artificial intelligence review 53, 8 (2020), 5455–5516.
Yann LeCun, Yoshua Bengio, and Geoffrey Hinton. 2015. Deep learning. nature 521, 7553 (2015), 436–444.
Yann LeCun, Bernhard Boser, John Denker, Donnie Henderson, Richard Howard, Wayne Hubbard, and Lawrence Jackel. 1989. Handwritten digit recognition with a back-propagation network. Advances in neural information processing systems 2 (1989).
Li Li, Miloš Doroslovački, and Murray H Loew. 2020. Approximating the gradient of cross-entropy loss function. IEEE access 8 (2020), 111626–111635.
Xiaofan Lin, Cong Zhao, and Wei Pan. 2017. Towards accurate binary convolutional neural network. Advances in neural information processing systems 30 (2017).
Majid Memari, Praveen Shakya, Mohammad Shekaramiz, Abdennour C Seibi, and Mohammad AS Masoum. 2024. Review on the advancements in wind turbine blade inspection: Integrating drone and deep learning technologies for enhanced defect detection. IEEE Access (2024).
Vinod Nair and Geoffrey E Hinton. 2010. Rectified linear units improve restricted boltzmann machines. In Proceedings of the 27th International Conference on Machine Learning (ICML-10). 807–814.
Vladimir Nasteski. 2017. An overview of the supervised machine learning methods. Horizons. b 4, 51-62 (2017), 56.
Otacílio de A Ramos Neto, Rafael C Chaves, Alysson P Nascimento, and Ruan D Gomes. 2024. Middleware para aplicações distribuídas de vídeo com suporte à computação na borda na Indústria 4.0. In Brazilian Symposium on Multimedia and the Web (WebMedia). SBC, 215–222.
Han Peng, Hai Zhang, Yisa Fan, Linjian Shangguan, and Yang Yang. 2022. A review of research on wind turbine bearings’ failure analysis and fault diagnosis. Lubricants 11, 1 (2022), 14.
David MW Powers. 2020. Evaluation: from precision, recall and F-measure to ROC, informedness, markedness and correlation. arXiv preprint arXiv:2010.16061 (2020).
Wei Qiao and Dingguo Lu. 2015. A survey on wind turbine condition monitoring and fault diagnosis—Part II: Signals and signal processing methods. IEEE Transactions on Industrial Electronics 62, 10 (2015), 6546–6557.
Christine Röckmann, Sander Lagerveld, and John Stavenuiter. 2017. Operation and maintenance costs of offshore wind farms and potential multi-use platforms in the Dutch North Sea. In Aquaculture Perspective of Multi-Use Sites in the Open Ocean: The Untapped Potential for Marine Resources in the Anthropocene. Springer International Publishing Cham, 97–113.
Takaya Saito and Marc Rehmsmeier. 2015. The precision-recall plot is more informative than the ROC plot when evaluating binary classifiers on imbalanced datasets. PloS one 10, 3 (2015), e0118432.
Marina Sokolova and Guy Lapalme. 2009. A systematic analysis of performance measures for classification tasks. Information processing & management 45, 4 (2009), 427–437.
Bouno Toshio, Yuji Toshifumi, Hamada Tsugio, and Hideaki Toya. 2005. Failure forecast diagnosis of small wind turbine using acoustic emission sensor. KIEE International Transaction on Electrical Machinery and Energy Conversion Systems 5, 1 (2005), 78–83.
Xin Wang, Dongxing Mao, and Xiaodong Li. 2021. Bearing fault diagnosis based on vibro-acoustic data fusion and 1D-CNN network. Measurement 173 (2021), 108518.
Khalid Zaman, Melike Sah, Cem Direkoglu, and Masashi Unoki. 2023. A survey of audio classification using deep learning. IEEE Access 11 (2023), 106620–106649.
Pinjia Zhang and Delong Lu. 2019. A survey of condition monitoring and fault diagnosis toward integrated O&M for wind turbines. Energies 12, 14 (2019), 2801.
Yang Zhang, Maciej Radzieński, Sidney Xue, Ziping Wang, and Wiesław Ostachowicz. 2024. A method for detecting cracks on the trailing edge of wind turbine blades based on aeroacoustic noise analysis. Structural Health Monitoring (2024), 14759217241303452.
Published
2025-11-10
How to Cite
OLIVEIRA, Gabriel Luiz Barros De; ALVES, Lucas; ROCHA, Cleilton L.; LIMA, João Pedro B.; MESQUITA, Paulo; TRAJANO, Alex.
End-to-End Deep Learning Approach for Wind Turbine Bearing Fault Detection From Acoustic Data. In: BRAZILIAN SYMPOSIUM ON MULTIMEDIA AND THE WEB (WEBMEDIA), 31. , 2025, Rio de Janeiro/RJ.
Anais [...].
Porto Alegre: Sociedade Brasileira de Computação,
2025
.
p. 202-210.
DOI: https://doi.org/10.5753/webmedia.2025.15198.
