基于SVD降噪和盲信号分离的滚动轴承故障诊断

Fault diagnosis of rolling bearings based on SVD denoising and blind signals separation

滚动轴承早期微弱故障特征信号往往淹没于系统噪声信号中而难于识别,奇异值分解技术(SVD)可以有效降低噪声水平,提高周期成分的提取能力,盲源分离技术可以分离故障源信号并提取故障特征。将奇异值分解技术和盲信号分离技术的优势应用于滚动轴承故障诊断,利用奇异值分解降噪特性消除系统信号中的混合噪声,对降噪后的信号通过盲信号分离技术进行盲源分离,提取出原始故障信号。数值仿真及实验结果表明,该方法可以成功地分离出滚动轴承实测信号的典型故障,提高滚动轴承故障诊断的效果。

Early fault signal of a rolling bearing submerged in system noises is weak and difficult to identify Singular value decomposition （SVD） technology can reduce the noise effectively and extract periodic components, blind signals separation（BSS） can separate fault source signals and extract fault characteristics. Here, using the advantages of SVD and BSS for rolling bearing fault diagnosis, the composite noise of system signals was eliminated with the SVD algorithm, the signals after processing were separated with the BSS algorithm, then the original fault signal characters were extracted. The numerical simulation examples and experimental results showed that this method is successful in separating typical faults for rolling bearings, it improves the efficiency of rolling bearing fault diagnosis.