Wind Turbine Acoustic Anomaly Detection based on RepViT-MobileNetV3

Qingzheng Li

doi:10.6919/ICJE.202512_11(12).0016

Authors

Qingzheng Li

DOI:

https://doi.org/10.6919/ICJE.202512_11(12).0016

Keywords:

Gearbox; Anomalous Sound Detection; MobileNetV3.

Abstract

Abnormal sound detection is a technique used to recognize non-normal sound signals, which is widely used in industrial fields, such as abnormal detection of wind turbines. Currently, many abnormal sound detection techniques are based on deep learning. However, in the complex working environment of WTGs, which is filled with a large amount of noise, abnormal sound detection for WTGs faces problems such as difficulty in sound feature extraction and insufficient sound feature extraction capability of the detection network. Therefore, this paper proposes a deep learning-based method for abnormal sound detection of WTGs, called RS-MobileNet.Specifically, SincNet spectral features and Log-Mel spectral features are extracted from the original sound signals, which are fused to become SL spectrograms as feature input. Then the improved RS-MobileNetV3 network is proposed based on the MobileNetV3 network. This network combines the reparameterized visual transform module and the soft pooling method, which can make the MobileNetV3 network keep lightweight while improving the feature extraction capability. Using the NREL dataset, the AUC is improved by 4.53 percentage points compared to the baseline model MobileNetV3.

Downloads

Download data is not yet available.

References

[1] Jiarui L ,Guotian Y ,Xinli L , et al.Wind turbine anomaly detection based on SCADA: A deep autoencoder enhanced by fault instances.[J].ISA transactions,2023,139586-605.

[2] Yida W ,Joseph D T ,Nassir N , et al.SoftPool++: An Encoder–Decoder Network for Point Cloud Completion[J].International Journal of Computer Vision,2022,130(5):1145-1164.

[3] Chunyuan W ,Yang W ,Yihan W , et al.Scene Recognition Using Deep Softpool Capsule Network Based on Residual Diverse Branch Block[J].Sensors,2021,21(16):5575-5575.

[4] Nantian H ,Qingzhu C ,Guowei C , et al.Fault Diagnosis of Bearing in Wind Turbine Gearbox Under Actual Operating Conditions Driven by Limited Data With Noise Labels[J].IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT,2021,70

[5] Ziqiang P ,Chuan L ,Shaohui Z , et al.Fault Diagnosis for Wind Turbine Gearboxes by Using Deep Enhanced Fusion Network[J].IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT,2021,70

[6] Renström N ,Bangalore P ,Highcock E .System-wide anomaly detection in wind turbines using deep autoencoders[J].Renewable Energy,2020,157(prepublish):647-659.

[7] Feng Z ,Zhu W ,Zhang D .Time-Frequency demodulation analysis via Vold-Kalman filter for wind turbine planetary gearbox fault diagnosis under nonstationary speeds[J].Mechanical Systems and Signal Processing,2019,12893-109.

[8] Cao L ,Qian Z ,Zareipour H , et al.Fault Diagnosis of Wind Turbine Gearbox Based on Deep Bi-Directional Long Short-Term Memory Under Time-Varying Non-Stationary Operating Conditions.[J].IEEE Access,2019,7155219-155228.

[9] Wang A, Chen H, Lin Z, et al. RepViT-SAM: Towards Real-Time Segmenting Anything[J]. arXiv preprint arXiv:2312.05760, 2023.

[10] 3TIER, NREL. Western Wind Integration Data Set[R]. National Renewable Energy Laboratory, 2022.