摘要
风电机组行星齿轮箱振动信号是一种非平稳、非线性信号,传统故障检测方法对于此类信号处理能力有限。将卷积自动编码器引入风机故障检测领域,构建了一种一维卷积自动编码器网络结构。首先训练卷积自动编码器无监督的提取数据特征,得到行星轮不同健康状况的特征向量,再对特征向量求取平均值获得指标向量。通过监督学习获得最优闵可夫斯基指数,最后通过测试数据的特征向量和指标向量之间的闵式距离来判断故障类型,实现了行星齿轮不同健康状况数据的识别和分类。实验结果证明,该方法可以有效的提取行星齿轮箱故障特征并达到诊断故障的目的。
The vibration signals of planetary gearbox of wind turbines are non-stationary and nonlinear signals. Traditional fault detection methods have limited capacity to process such signals. In this paper, convolutional automatic encoder is introduced into the field of wind turbine fault detection, and a network structure of one dimensional convolutional automatic encoder is constructed. Firstly, unsupervised extraction of data features is performed using convolutional automatic encoders. The characteristic vectors and index vectors of different health conditions of planetary gear are obtained. Optimal Minkowski index is obtained by supervised learning, and the fault type is determined by Minkowski distance between the feature vector of the test data and the index vector. Experimental results show that the method can effectively extract the fault characteristics of planetary gear box and achieve fault diagnosis purpose.
作者
李东东
王浩
杨帆
郑小霞
华伟
邹胜华
LI Dongdong;WANG Hao;YANG Fan;ZHENG Xiaoxia;HUA Wei;ZOU Shenghua(School of Electrical Engineering,Shanghai University of Electric Power,Yangpu District,Shanghai 200090,China;School of Automation Engineering,Shanghai University of Electric Power,Yangpu District,Shanghai 200090,China;State Grid Northeast of Jiangxi Power Supply Branch,Leping 333300,Jiangxi Province,China)
出处
《电网技术》
EI
CSCD
北大核心
2018年第11期3805-3811,共7页
Power System Technology
基金
国家自然科学基金项目(51407114,51507098)
上海市科学技术委员会资助(13DZ2251900,10DZ2273400)
上海市“曙光计划”项目(15SG50)~~
关键词
行星齿轮箱
卷积自动编码器
闵式距离
无监督学习
故障诊断
planetary gearbox
convolutional automatic encoder
Minkowski distance
unsupervised learning
fault diagnosis