Research And Application On Lmd In The Rotating Machinery Vibration

As a time-frequency method, local mean decomposition(LMD) appeared in recentyears and can process signal adaptively. To the nonlinear and non-stationary signalgenerated in rotating machinery vibration, this signal processing method shows goodeffect, so it provides a strongly theoretical basis for the mechanical condition monitoringand repair.This paper analyses the characteristics of vibration signal of rotating machinery fault,and using the rotor test bench simulates the mechanical fault. According to some problemspossibly created in the decomposition process, this paper analyzes the endpoints, movingaverage span and the loop-termination criteria, and proposes new endpoint processingmethod and loop-termination criteria. In view of empirical mode decomposition(EMD)removing the cross-term of the WVD, this paper analyzes using LMD to remove thecross-term, and achieves good results. According to getting the component using theoriginal LMD has low accuracy, this paper proposes LMD based on decomposing thecomponent again. Then, LMD based on steady point is put forward, and this kind ofimproved method improves the accuracy and the speed of decomposition. Through theanalysis of the simulated signal and practical signal, the results show the improvement forthe original LMD in various aspects is effective.Finally, the improved LMD is applied to the practical engineering. The results ofcomparing the improved method with HG8904C fault diagnosis system，shows theimproved method is effective and practical and can be applied to the rotating machineryfault diagnosis.