Research on wind turbine blade fault detection based on DenseNet-TL combined with ELM

Aiming at the safety hidden danger caused by blade faults that are difficult to be detected, the fault prevention detection technology based on blade image intelligent processing is investigated. A wind turbine blade fault prevention detection method is designed to analyses wind turbine blade images...

Full description

Saved in:
Bibliographic Details
Main Authors: Dianming WANG, Xue PAN, Jian MA, Yuzhang DAI, Chengjun SUN, Shibin LI, Xiaoju YIN
Format: Article
Language:English
Published: Tamkang University Press 2025-05-01
Series:Journal of Applied Science and Engineering
Subjects:
Online Access:http://jase.tku.edu.tw/articles/jase-202512-28-12-0012
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849714398880333824
author Dianming WANG
Xue PAN
Jian MA
Yuzhang DAI
Chengjun SUN
Shibin LI
Xiaoju YIN
author_facet Dianming WANG
Xue PAN
Jian MA
Yuzhang DAI
Chengjun SUN
Shibin LI
Xiaoju YIN
author_sort Dianming WANG
collection DOAJ
description Aiming at the safety hidden danger caused by blade faults that are difficult to be detected, the fault prevention detection technology based on blade image intelligent processing is investigated. A wind turbine blade fault prevention detection method is designed to analyses wind turbine blade images by combining DenseNet, Transfer Learning (TL) and Extreme Learning Machines (ELM), and collect image samples as a training set. The image samples are collected as training set, and the image features are effectively extracted using the improved DenseNet, which is combined with Extreme Learning Machines to improve the classification accuracy of the detection. 8000 images were collected, and the analysis results for the test set of images show that the detection accuracy of this model is higher than that of the DenseNet, ResNet and AlexNet models of migration learning, reaching more than 99%, and obtaining a more accurate preventive detection of wind turbine blade faults.
format Article
id doaj-art-e792ea2c8b1d482295ce33207c4e7d0b
institution DOAJ
issn 2708-9967
2708-9975
language English
publishDate 2025-05-01
publisher Tamkang University Press
record_format Article
series Journal of Applied Science and Engineering
spelling doaj-art-e792ea2c8b1d482295ce33207c4e7d0b2025-08-20T03:13:42ZengTamkang University PressJournal of Applied Science and Engineering2708-99672708-99752025-05-0128122451246010.6180/jase.202512_28(12).0012Research on wind turbine blade fault detection based on DenseNet-TL combined with ELMDianming WANG0Xue PAN1Jian MA2Yuzhang DAI3Chengjun SUN4Shibin LI5Xiaoju YIN6School of Renewable Energy, Shenyang Institute of Engineering, Shenyang 110136, Liaoning Province, ChinaSchool of Renewable Energy, Shenyang Institute of Engineering, Shenyang 110136, Liaoning Province, ChinaHuaneng International Power Company Limited Dandong Power Plant, Donggang 110167, Liaoning Province, ChinaHuaneng International Power Company Limited Dandong Power Plant, Donggang 110167, Liaoning Province, ChinaHuaneng International Power Company Limited Dandong Power Plant, Donggang 110167, Liaoning Province, ChinaHuaneng International Power Company Limited Dandong Power Plant, Donggang 110167, Liaoning Province, ChinaSchool of Renewable Energy, Shenyang Institute of Engineering, Shenyang 110136, Liaoning Province, ChinaAiming at the safety hidden danger caused by blade faults that are difficult to be detected, the fault prevention detection technology based on blade image intelligent processing is investigated. A wind turbine blade fault prevention detection method is designed to analyses wind turbine blade images by combining DenseNet, Transfer Learning (TL) and Extreme Learning Machines (ELM), and collect image samples as a training set. The image samples are collected as training set, and the image features are effectively extracted using the improved DenseNet, which is combined with Extreme Learning Machines to improve the classification accuracy of the detection. 8000 images were collected, and the analysis results for the test set of images show that the detection accuracy of this model is higher than that of the DenseNet, ResNet and AlexNet models of migration learning, reaching more than 99%, and obtaining a more accurate preventive detection of wind turbine blade faults.http://jase.tku.edu.tw/articles/jase-202512-28-12-0012wind turbine bladeextreme learning machinetransfer learningimage recognitiondensenet
spellingShingle Dianming WANG
Xue PAN
Jian MA
Yuzhang DAI
Chengjun SUN
Shibin LI
Xiaoju YIN
Research on wind turbine blade fault detection based on DenseNet-TL combined with ELM
Journal of Applied Science and Engineering
wind turbine blade
extreme learning machine
transfer learning
image recognition
densenet
title Research on wind turbine blade fault detection based on DenseNet-TL combined with ELM
title_full Research on wind turbine blade fault detection based on DenseNet-TL combined with ELM
title_fullStr Research on wind turbine blade fault detection based on DenseNet-TL combined with ELM
title_full_unstemmed Research on wind turbine blade fault detection based on DenseNet-TL combined with ELM
title_short Research on wind turbine blade fault detection based on DenseNet-TL combined with ELM
title_sort research on wind turbine blade fault detection based on densenet tl combined with elm
topic wind turbine blade
extreme learning machine
transfer learning
image recognition
densenet
url http://jase.tku.edu.tw/articles/jase-202512-28-12-0012
work_keys_str_mv AT dianmingwang researchonwindturbinebladefaultdetectionbasedondensenettlcombinedwithelm
AT xuepan researchonwindturbinebladefaultdetectionbasedondensenettlcombinedwithelm
AT jianma researchonwindturbinebladefaultdetectionbasedondensenettlcombinedwithelm
AT yuzhangdai researchonwindturbinebladefaultdetectionbasedondensenettlcombinedwithelm
AT chengjunsun researchonwindturbinebladefaultdetectionbasedondensenettlcombinedwithelm
AT shibinli researchonwindturbinebladefaultdetectionbasedondensenettlcombinedwithelm
AT xiaojuyin researchonwindturbinebladefaultdetectionbasedondensenettlcombinedwithelm