Condition monitoring of bearings based on optimal weight impulse extraction

The periodic impulses resulted from the faults of bearings are inevitably affected by noise, which may lead the fault difficult to be detected from monitoring signal. The commonly used signal processing methods aim to solve a filter through deconvolution. However, these methods fail to address the issue of noise interference on deconvolution process. To overcome this limitation, this paper proposed an optimal weight impulse extraction (OWIE) method to suppress the noise interference and highlight impulse components of monitoring signals. First, a weight sequence is utilized to preserve the non-impulse part of the raw signal while suppressing the impulse components. Then, by subtracting the weighted signal from the raw signal, the impulse components of signal can be highlighted. An iterative procedure is designed to solve the local optimal solution of the weight sequence by maximizing the kurtosis of impulse signal. The effectiveness of the proposed OWIE method is validated through a simulation of bearing fault signals and a case study of bearing run-to-failure dataset. The results demonstrate that the OWIE is capable of extracting periodic impulse components from monitoring signals and accurately distinguishing the health state of machines.