Detection And Modeling Of Gear Fault Based On Alpha Stable Distribution

Gears, as the most commonly used parts of machinery and the most vulnerableparts to failure of the gearbox, whether gears work well can directly influence therunning state of the whole machine. Therefore, detection of the gear fault in thestudy of gears is very necessary.The Alpha stable distribution which can be used to describe the signals withobvious pulse feature accurately by analyzing the non-Gaussian characteristics andthe pulse characteristics in simulation fault signal of gears as well as the actual gearfault signals. Alpha stable distribution model is generalized gaussian distributionmodel and it has higher accuracy in fitting gear fault signal probability densitydistribution aspects, and at the same time also more reasonable. For the first time,gear simulation signal was proved that it is consistent with Alpha stable distributionby the definition and properties of the distribution in this paper, which provides thetheoretical foundation for the use of Alpha stable distribution model.Then, by analyzing the influence of the parameters of Alpha stable distributionand kurtosis which were caused by the gear fault level and other operating conditongsuch as speed and noise. Then a evaluation method KLD was proposed based on theparameters of Alpha stable distribution, a new evaluation method Lwas proposedbased on KLD, through simulation and experimental analysis, we compared it withthe KLD, the analysis results show that the method of L can detect the gear earlierand it is sensitivity to failure pulse, but the later evaluation effect of L is worse thanKLD.At last, combined with adaptive wavelet, an early-staged fault signal detectionmethod was put forward based on the parameters of Alpha stable distributionparameters, simulation and experimental results indicates that this method canachieve good effect compared with the detection method based on adaptive waveletcombined with kurtosis. This adaptive wavelet detection method based on the Alphastable distribution parameters has practical application value, it can detect the gearfault signal earlier and avoid the failure cause of the accident and loss.