摘要
在基于奇异值分解技术的信号处理中,针对齿轮振动信号中微弱故障特征难以提取的问题,提出了一种选择有效奇异值的方法。该方法对全部的奇异值单独重构,得到对应的一维信号并对其进行频谱分析,结合齿轮箱振动信号的特征频率成分,找出对应齿轮频谱特征的奇异值,将其进行重构得到最终经SVD处理的信号。试验处理结果表明,此法可有效提取出齿轮裂纹信号中的微弱特征,同时该方法可广泛应用于旋转机械信号中微弱特征成分的提取。
For the signal processing based on singular value decomposition, the faint fault feature is hard to be ex- tracted from the gear vibration signals. A selection method of effective singular values is proposed here to solve problem. By using every separate singular value, the one-dimensional signals are reconstructed. The frequency spectrums of the reconstructed signals are gained through Fast Fourier Transformation （FFT）. Chose the singular values which correspond with the feature frequency in the gear signal; and use them to reconstruct the signal as the ultimately result. The procession results show that this method can extract the faint fault feature from the gear crack fault signal and it can be widely used in the fault feature extraction of the rotating machinery.
出处
《机械科学与技术》
CSCD
北大核心
2012年第9期1449-1453,共5页
Mechanical Science and Technology for Aerospace Engineering
关键词
奇异值分解
振动信号
故障特征
有效奇异值
频谱分析
singular value decomposition
vibration signal
fault feature
effective singular value
frequency analysis
