Abstract:Due to practical data were easily interrupted and useful information was covered by noises, a novel de-noising approach was proposed based on the adaptive structure element for generalized morphological filtering (ASEGMF). Firstly, the sine structure element was selected according to the feature of vibration signal, and the length scale and the height scale were defined. Secondly, the peak distance and the peak height were defined by signal’s local characteristic, and the length scale and the height scale of sine structure element were gotten by adaptive method. Finally, the interrupted vibration signal was de-noised by the generalized morphological filter cascaded by one small and one big adaptive structure elements. This method conquers the selective random of current morphological filter, the structure element are gotten adaptively by signal’s local characteristic without artificial interference. Practical and simulation results show that this method has better de-nosing effectiveness. It’s suitable for on-line monitoring and diagnosis of rotating machinery.