The Wavelet Transform in the Acoustic Emission Signal Feature Extraction of the Rubbing Fault

The rubbing fault of a rotary machine was simulated with a rubbing-block made of 0Cr18Ni9 steel at the rate of 310 r.p.m. The acoustic emission signals were sampled for the case with rubbing-block and the case without rubbing-block respectively. The feature extraction of acoustic emission signals was studied by means of db10 function of the wavelet transform and the wavelet energy. It is shown that there are 3 kinds of waveforms for the signals from the experiment with a rubbing block, the feature frequency ranges of the rotor rubbing are about 830-920 KHZ at d1 level and about 235-385 KHZ at d3 level, and the energy concentrates on the decomposed signals at d2 and d3 levels for the case with rotor rubbing. The energy ratio feature makes the recognition of the rotor rubbing fault easier.