基于奇异值分解和S变换的齿轮故障诊断研究

Research on Gear Fault Diagnosis Based on Singular Value Decomposition and S-transform

为了在强背景噪声干扰下能从核电行星齿轮箱齿轮故障振动信号中提取出冲击特征进行故障诊断,文章将奇异值分解与S变换相结合,并借助MATLAB软件辅助来处理信号中的噪声,提取突出故障信息的冲击特征。采用的方法为:从含有噪声的信号中建立Hankel矩阵,使用3种不同的奇异值选择方法,即奇异值中值法、特征均值法以及奇异值差分谱法,对Hankel矩阵进行奇异值处理后,再重构信号,达到降低信号噪声的目的;对重构信号进行S变换,获得该重构信号的时频谱图;对时频谱图进行S逆变换,得到重构信号的冲击特征。研究结果表明:将奇异值差分谱法与S变换相结合提取出冲击特征的效果最佳,可以为核电行星齿轮箱齿轮相关故障的诊断提供先验信息。

In order to extract the impact feature from the vibration signal of the gear fault of the nuclear power planetary gear box for fault diagnosis under the interference of strong background noise,this paper combines the Singular Value Decomposition(SVD)and S transform,and uses MATLAB software to deal with the noise in the signal and extract the impact feature of the prominent fault information.The method adopted is as follows:the Hankel matrix is established from the signal containing noise,and three different singular value selection methods are used,namely,singular value median method,characteristic mean method and singular value difference spectrum method.After singular value processing,the Hankel matrix is reconstructed to achieve the purpose of reducing signal noise.The time spectrum diagram of the reconstructed signal is obtained by S-transform.The impact characteristics of the reconstructed signal are obtained by S inverse transformation of the time spectrum diagram.The results show that the combination of singular value difference spectrum method and S-transform is the best method to extract the impact characteristics,which can provide prior information for the diagnosis of gear related faults of nuclear power planetary gearbox.