摘要
针对带式输送机托辊轴承故障诊断中振动信号提取特征困难而导致故障诊断精度较低的难题,提出了一种基于一维卷积神经网络(1DCNN)和极限学习机(ELM)的托辊轴承故障诊断方法。首先,根据具体的故障诊断任务,对采集到的数据进行划分,并进行傅里叶变换。然后,利用1DCNN提取故障特征,根据提取的故障特征利用ELM进行故障分类。测试精度均达到100%,用时分别为2.82和2.42 s。通过与ELM、随机森林、K最邻近法、支持向量机和卷积神经网络等方法对比,体现了所提方法的优越性。结果表明:采用1DCNN和ELM相结合的诊断方法,其诊断效果相比单一方法更好,能够满足煤矿领域托辊故障诊断需求。
Aiming at the problem of low fault diagnosis accuracy caused by the difficulty of vibration signal extraction in the fault diagnosis of belt conveyor roller bearing,a fault diagnosis method of roller bearing based on one-dimensional convolutional neural network(1DCNN)and extreme learning machine(ELM)was proposed.Firstly,according to the specific fault diagnosis task,the collected data was divided and Fourier transform was performed.Then,1DCNN was used to extract fault features,and ELM was used to classify faults according to the extracted fault features.The test accuracy reaches 100%,and the test time is 2.82 s and 2.42 s,respectively.By comparing with ELM,random forest,K-nearest neighbor method,support vector machine and convolutional neural network,the superiority of the proposed method were demonstrated.The results show that the diagnosis method combining 1DCNN and ELM is better than the single method,which can meet the needs of roller fault diagnosis in coal mine field.
作者
赵亚东
马腾飞
胡小刚
武轶凡
ZHAO Yadong;MA Tengfei;HU Xiaogang;WU Yifan(Bulianta Coal Mine,Guoneng Shendong Coal Group Co.,Ltd.,Ordos 017200,China;China Coal Technology&Engineering Group Information Co.,Ltd.,Xi′an 710000,China)
出处
《洁净煤技术》
CAS
CSCD
北大核心
2024年第S01期587-592,共6页
Clean Coal Technology
关键词
一维卷积神经网络
极限学习机
托辊
轴承
故障诊断
one-dimensional convolutional neural network
extreme learning machine
roller
bearings
fault diagnosis
作者简介
赵亚东(1985—),男,陕西榆林人,工程师,硕士。E-mail:zyd5102@163.com
