Truncation filtering method for envelope analysis

Summary Envelope analysis is an effective tool for bearing fault diagnosis. In most cases, the fault signal of the bearing that is located deeply inside the machine is always covered by the huge wide‐band noise. However, the common envelope analysis cannot effectively extract the fault feature from the frequency band containing huge wide‐band noise. Therefore, a modified envelope analysis that contains a new band‐pass filter based on resonance band and a new matched filtering computing method is proposed. The common envelope analysis is always concentrated on the selection of the filtering band and filter's other parameters, which can just control some parts of the filter. However, the whole shape of the filter has great effects on every element content of the filtered signal. Thus, according to the distribution characteristic of fault impact spectrum lines, the shape of the freedom damping vibration excited by single impact in frequency domain can be used as a filter for the truncation filtering to obtain the optimal envelope spectrum. Since the envelope shape of practical fault impact spectrum lines is not exactly like the shape of single impact in frequency domain, a scan filtering method is added to the band‐pass filtering, which is to use different central frequencies with the same interval and different bandwidths for filtering. A complicated structure test rig that is self‐innovate is used for experimental validation. The experimental result shows that fault feature extracted by the proposed method is better than the one extracted by the common envelope analysis.