摘要
基于神经网络的滚动轴承故障诊断方法训练时,存在诊断准确率低和易受到变工况噪声干扰的问题,提出一种基于重参数化VGG(RepVGG)滚动轴承故障诊断方法。为满足神经网络对数据量的要求,采用数据增强技术来扩充原始数据,使用短时傅里叶变换(STFT)对原始的振动信号处理成单通道时频图,并使用伪彩色处理技术转换成三通道时频图,进一步将数据输入到RepVGG网络的不同结构中进行滚动轴承的故障诊断。在凯斯西储大学(CWRU)滚动轴承数据集上开展试验验证,试验结果表明,RepVGG在变工况及噪声干扰下的平均诊断准确率分别为98.02%、95%以上,高于基于VGG、ResNet的故障诊断模型,有较高的故障诊断准确率且泛化性更好。
When training neural network based rolling bearing fault diagnosis method,there are problems of low diagnostic accuracy and susceptibility to interference from variable operating noise.A rolling bearing fault diagnosis method based on reparameterized VGG(RepVGG)was proposed to solve these problems.Firstly,to meet the data volume requirement of the neural network,the data enhancement technique was used to expand the original data.Then,the original vibration signal was processed into a single-channel time-frequency map using the short-time Fourier transform(STFT)and converted into a three-channel time-frequency map using the pseudo-color processing technique,and the data were further input into different structures of the RepVGG network for the fault diagnosis of rolling bearings.Finally,experimental validation was carried out on the rolling bearing dataset of Case Western Reserve University(CWRU),and the experimental results show that the average diagnostic accuracy of RepVGG under variable operating conditions and noise interference is 98.02%and over 95%,respectively,which is higher than that of VGG and ResNet fault diagnosis models,with higher fault diagnosis accuracy and stronger generalization.
作者
丁汕汕
陈仁文
黄翊君
刘飞
刘昊
肖安
DING Shanshan;CHEN Renwen;HUANG Yijun;LIU Fei;LIU Hao;XIAO An(State Key Laboratory of Mechanics and Control of Mechanical Structures,Nanjing University of Aeronautics and Astronautics,Nanjing 210016,China;College of Automation Engineering,Nanjing University of Aeronautics and Astronautics,Nanjing 210016,China)
出处
《振动与冲击》
EI
CSCD
北大核心
2023年第11期313-323,共11页
Journal of Vibration and Shock
基金
国家自然科学基金项目(51635008)
江苏省高校优势学科建设工程项目(PAPD)。
作者简介
第一作者:丁汕汕,男,博士生,1994年生;通信作者:陈仁文,男,博士,教授,1966年生。
