低信噪比环境下圆锥滚子轴承微弱故障信号自动化诊断  

Automated diagnosis of weak fault signals in tapered roller bearings under low signal-to-noise ratio environments

作　　者：张国鸣 张志文 苏和 ZHANG Guoming;ZHANG Zhiwen;SU He(Storage and Transportation Center,Guoneng Beidian Shengli Energy Co.,Ltd.Inner Mongolia,Xilinhot 026000 China)

机构地区：[1]国能北电胜利能源有限公司储运中心,内蒙古锡林浩特026000

出　　处：《自动化与仪器仪表》2025年第3期121-125,共5页Automation & Instrumentation

基　　金：内蒙古自治区科技计划项目(2020GG023)。

摘　　要：低信噪比环境中的圆锥滚子轴承早期微弱故障识别,是轴承故障诊断领域的一个难点问题,当前基于希尔伯特变换解调的包络谱分析法虽是得到广泛工程应用的轴承故障检测经典方法,但其对信噪比过低的轴承早期微弱故障特征识别能力不足,为了及早发现圆锥滚子轴承异常情况,保障稳定运行,提出一种低信噪比环境下圆锥滚子轴承微弱故障信号自动化诊断方法。采用粒子群算法优化小波阈值函数,提取振动时频域高阶统计量——谱峭度特征。利用CNN和BiGRU双通道模型分析轴承微弱信号的谱峭度特征,通过改进压缩激励网络加权融合识别的轴承微弱信号谱峭度。根据已知故障样本和谱峭度特征的加权融合结果,分类新的振动信号,实现微弱故障信号自动化诊断。实验结果表明,所提方法可以精准诊断圆锥滚子轴承微弱故障。The early weak fault identification of tapered roller bearings in low signal-to-noise ratio environments is a difficult problem in the field of bearing fault diagnosis.Although the envelope spectrum analysis method based on Hilbert transform demodulation is a widely used classic method for bearing fault detection in engineering,its ability to identify early weak fault characteristics of bearings with low signal-to-noise ratio is insufficient.In order to detect abnormal situations of tapered roller bearings early and ensure stable operation,Propose an automated diagnostic method for weak fault signals of tapered roller bearings in low signal-to-noise ratio environments.Using particle swarm optimization algorithm to optimize wavelet threshold function and extract high-order statistics in vibration time-frequency domain-spectral kurtosis features.Using CNN and BiGRU dual channel models to analyze the spectral kurtosis characteristics of weak signals in bearings,and improving the compression excitation network weighted fusion recognition of bearing weak signal spectral kurtosis.Based on the weighted fusion results of known fault samples and spectral kurtosis features,new vibration signals are classified to achieve automated diagnosis of weak fault signals.The experimental results show that the proposed method can accurately diagnose weak faults in tapered roller bearings.

关 键 词：圆锥滚子轴承 微弱故障信号 谱峭度 粒子群算法 卷积神经网络 

分 类 号：TH133.3[机械工程—机械制造及自动化]