Wind turbine bearing defect identification method based on improved SOLOv2

Xuecun Yang; Liyuan Chen

doi:10.1117/12.2689796

10 August 2023 Wind turbine bearing defect identification method based on improved SOLOv2

Xuecun Yang, Liyuan Chen

Proceedings Volume 12748, 5th International Conference on Information Science, Electrical, and Automation Engineering (ISEAE 2023); 127480J (2023) https://doi.org/10.1117/12.2689796
Event: 5th International Conference on Information Science, Electrical and Automation Engineering (ISEAE 2023), 2023, Wuhan, China

Abstract

Aiming at the problems of low recognition rate of complex shape defects on wind turbine bearing surface and inaccurate localization of small target defects by traditional inspection methods, this paper proposes an improved SOLOv2-based bearing surface defect recognition method. The method selects ResNeXt-101-FPN as the backbone network on the basis of SOLOv2 model framework, combines deformable convolution (DCNv2) to strengthen the learning of defect morphological features by the network, which is more adaptable to complex defect objects; proposes improved mask feature branching, adds adaptive attention module and feature enhancement module, which not only reduces the loss of feature information, but also enhances the feature expression and change the way of feature fusion. Experiments show that the mean average precision (mAP) of the improved SOLOv2 network for bearing surface defect recognition is 96.6%, which is 2.8% higher than that before the improvement. The average recall (AR) is 96.8%, which is 1.6% higher than before the improvement; the improved model is less affected by the variable shape of the defects, locates small targets more accurately, and effectively solves the problems of partial missed segmentation and under-segmentation.

Citation Download Citation

Xuecun Yang and Liyuan Chen "Wind turbine bearing defect identification method based on improved SOLOv2", Proc. SPIE 12748, 5th International Conference on Information Science, Electrical, and Automation Engineering (ISEAE 2023), 127480J (10 August 2023); https://doi.org/10.1117/12.2689796

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available