Abstract: Vibration signals of rotating machinery acquired on the field are usually accompanied with impulse and white noise.Wavelet threshold de-noising is effective for filtering white noise but ineffective in filtering impulse noise,and morphological filter can effectively remove impulse noise but it is difficult for filtering white noise.An integrated filter based on morphological de-noising and wavelet threshold de-noising was proposed to efficiently purify the corrupted vibration signals of rotating machinery.The proposed filter algorithm presents the combined advantages of both morphological and wavelet filters,and it can simultaneously filter both impulse and white noises in the rotor vibration signal.Based on the vibration signals of rotating machinery acquired in field and simulation results,the integrated filter which is composed of a morphological filter followed by an improved wavelet threshold filter presents a decent de-noising result,and extracts the rotor vibration signal submerged in the noise.The efficiency of the filter is experimentally demonstrated compared with the previously reported filters.

Key words: Signal processing,Morphological filter,Wavelet threshold filtering,Morphological-wavelet threshold filter