Jixie chuandong (Jan 2018)
Research of the Gear Fault Diagnosis based on Improved LMD and Manifold Learning
Abstract
In order to diagnosis gear fault efficiently by using vibration signal,a new method of gear fault based on local mean decomposition（LMD）,fuzzy entropy and Isomap extraction is proposed,this method combines LMD,fuzzy entropy and Isomap. Firstly,by using the local mean decomposition（LMD） to decomposed the original vibration signal to obtain the components in different scales,and increases the adaptive matching waveform to alleviate the influence of end effects on decomposition results in the original LMD method. Then,considering fuzzy entropy can be use to distinguish the complexity of the signal effectively,so the fuzzy entropy of Product functions（PF） by LMD is calculated,a high-dimensional feature vector can be obtain with product functions. Finally,by using manifold learning（ISOMAP） on the high dimensional feature into low dimensional features which have better discrimination to describe different gear fault. It is applied to the gear experiment,the experimental results show that the method can effectively diagnose the gear faults and has certain superiority.
