Fault feature extraction of rolling element bearings based on short-time processing

Fault diagnosis of bearings is a crucial part the maintenance process rotary machinery. Extracting cyclic characteristics impact force significant importance for bearing diagnosis. To highlight fault features from signals combined with heavy background noise, novel approach based on short-time processing proposed. are regarded as periodic impulse response signals. Firstly, vibration signal band-pass filtered subsequent spectral analysis. Then we integrate energy constant length, and natural logarithm considered to obtain curve. The curve straight decaying curve, its more concentrated characteristic frequency compared envelope. Finally, found by analysis effectiveness proposed method verified simulation experiments. harmonics sidebands in logarithmic spectrum suppressed well, highlighted. Comparison Hilbert envelope shows that can frequency.