Study On Application Of Multi-scale Morphology In Vibration Signal Processing

Vibration signal monitoring is widely used in condition monitoring and faultdiagnosis of mechanical equipment,and the vibration signal contains a wealth ofinformation about fault features. Accurately extract feature information are important tojudge the equipment status. Mathematical Morphology is a new non-linear signalprocessing method in time domain, whose calculation speed is faster than traditionalmethods, and the effect is well. So it’s more suitable for real-time monitoring of vibrationsignals. Multiscale morphology is a branch of mathematical morphology, which canachieve multi-resolution signal analysis.This paper will application multi-scale morphological analysis into vibration signalprocessing.The method is validated by theoretical analysis and experiment.Firstly introduced the application of mathematical morphology and research statusand the basic theory of mathematical morphology in the vibration signal processing.Thefrequency response characteristics of the multi-scale morphology is studied. Quantitativeanalysis of the effects of morphological parameters on the frequency responsecharacteristics. The length and amplitude of the structure elements, the samplingfrequency were studied separately,to provide theoretical support and guidance forpractical application.Secondly, the calculation process and mechanism of different morphologicaloperators were studied.Using geometry analysis shows visually the various operatorsmechanism, revealing the different operational characteristics of the operator, based onthat, selection method of structure elements and operator which used in vibration signalprocessing were proposedFinally, according to the research, we propose a new morphological vibration signalprocessing methods. Through simulation test rig gear boxes and bearing failure dataanalysis, feature extraction to achieve a failure to verify the feasibility and effectivenessof this method.