《机电工程》杂志社科技期刊核心期刊论文投稿机电一体化机械、仪表电工自动化、计算机浙江大学主办

Introduction

International Standard Serial Number:

ISSN 1001-4551

Sponsor:

Zhejiang University;

Zhejiang Machinery and Electrical Group

Edited by:

Editorial of Journal of Mechanical & Electrical Engineering

Chief Editor:

ZHAO Qun

Vice Chief Editor:

TANG ren-zhong,

LUO Xiang-yang

Tel:

86-571-87041360,87239525

Fax:

86-571-87239571

Add:

No.9 Gaoguannong,Daxue Road,Hangzhou,China

P.C:

310009

E-mail:

meem_contribute@163.com

Wind turbine bearing fault diagnosis method based on multi-source information fusion

Published:2023-10-30 author:WANG Zhengqi, GU Yanling, CHEN Changzheng, et al. Browse: 299 Check PDF documents

Wind turbine bearing fault diagnosis method based on multi-source information fusion

WANG Zhengqi1, GU Yanling1,2, CHEN Changzheng1,2, TIAN Miao1, SUN Xianming3

(1.School of Mechanical Engineering, Shenyang University of Technology, Shenyang 110870, China;

2.Liaoning Vibration and Noise Control

Engineering Research Center, Shenyang 110870, China;

3.Ningbo Kunbo Measurement and Control Technology Co., Ltd., Ningbo 315200, China)

Abstract: In the fault diagnosis of wind turbine rolling bearing, the information provided by a single sensor is limited. To solve this problem, in order to extract the multi-scale features of the input original signal to ensure the validity and integrity of the fault information, and to improve the efficiency and effectiveness of information fusion, a fault diagnosis method for wind turbine rolling bearings based on multi-source information fusion attention mechanism convolutional neural network (MSIF-ACNN) was proposed. Firstly, a fusion convolution method combining ordinary convolution and dilated convolution was proposed to extract multi-scale features from the original time-domain signal. Then, a two-layer channel and a spatial attention mechanism were adaptively used to calibrate and assign weights to different channel data. The multi-source information features output from the attention mechanism was fused. Finally, in order to verify the effectiveness of the multisource information fusion method, the classification method composed of full connection layer and classification layer was used to verify the actual bearing data of wind turbines. The experimental results show that there are huge differences in the sensitivity of sensors with different locations and different orientations for different rolling bearing faults. However, MSIF-ACNN effectively exploits this difference from these sensors and achieves complementarity between multisource information features. The MSIF-ACNN model achieves an accuracy of 96.7%, and its effect is better than other multi-source information diagnosis models. The proposed method promotes the application of information fusion in the field of wind turbine rolling bearing fault diagnosis.

Key words: wind turbine; rolling bearing; multi-scale features extraction; fault information integrity; multi-source information fusion(MSIF); attention mechanism convolutional neural network(ACNN)

Advanced Search

Online Office System

Chinese Core Periodicals
Chinese Sci-tech Core Periodicals
SA, INSPEC Indexed
CSA: T Indexed
UPD:Indexed