探测信号中周期性冲击分量的奇异值分解技术

机械设备振动信号中是否存在周期性的冲击分量是其有无故障的重要标志。通常检测的机械振动加速度信号 ,由于信噪比太低 ,即使存在周期性的冲击分量也往往被淹没在强的背景噪声之中。通过时域波形和频谱等基本分析手段来探测振动加速度信号中的周期性冲击分量往往是困难的。本文在总结基于奇异值分解的信号周期分量探测原理的基础上 ,针对现有信号奇异值分解技术存在的问题 ,对信号奇异值分解矩阵的构造方法作了重大改进。通过应用实例显示 ,用改进后的信号奇异值分解技术探测振动加速度信号中的周期性冲击分量是可行的

Detection of Periodic Impulse Components in Signals Using Singular Value Decomposition

Whether there exists a periodic impulse component in a vibration signal is an important sign of a mechanical device being faulty. However, in respect of a vibration acceleration signal, the signal to noise ratio (SNR) is often so small that the periodic impulse component is submersed in its intense background noises. It is difficult to detect the periodic impulse component in the vibration acceleration signal with conventional analysis methods such as time domain wave and frequency spectrum. On the basis of summarizing the principle of detecting the periodic component in the signal using singular value decomposition (SVD), this paper first finds out the disadvantage of the current detecting technique using SVD. Subsequently, the approach of constructing the matrix of SVD on the signal is improved largely. At last, a real life application example of detecting the periodic impulse component from a vibration acceleration signal is given, which shows that the improved SVD technique is feasible.