Application Of Improved Local Mean Decomposition Method In Gear Fault Diagnosis

With the increase of the proportion of the gear in the rotating machinery,the research of the gear becomes more and more important.As one of the important original gear transmission,in high load conditions,prone to failure,affecting the normal production process,which will lead to paralysis of the entire system,raise security issues and bring huge economic losses.The main failure forms of the gear are:wear,broken teeth,pitting,scuffing and so on.When the fault occurs,the amplitude and phase of the vibration signal will be changed.The accurate demodulation of vibration signal is a key point in the research of gear fault.In vibration signal processing,local mean decomposition method can decompose multi-component signals into multiple time domain components and frequency components,these components can get the complete recombinant signal time-frequency distribution,processing is very suitable for nonlinear and non-stationary signal.However,when the vibration signal is too strong or the signal length is too large,it will affect the calculation of the local mean decomposition,and even bring error to the diagnosis results.Therefore,based on the use of local mean decomposition method for fault diagnosis,combining with the Calman filter,said the support vector machine method and sparse compression advantage,noise reduction,classification and fault signals of vibration signal,in order to improve the accuracy of gear fault diagnosis,fault recognition rate and shorten the time of diagnosis.The main conclusions of this paper are as follows:(1)The vibration signal contains a lot of noise,which can easily affect the result of fault diagnosis.The local gear fault diagnosis method based on Calman filter,the vibration signal de-noising,spectrum comparison before and after noise reduction,noise reduction can be seen after the fault feature number from 1 to three,and the characteristic frequency is more obvious,which shows that the method can effectively reduce the impact of noise on the gear signal diagnosis.(2)The identification of gear fault type plays an important role in fault diagnosis.Based on local mean decomposition and support vector machine(SVM),the classification of gear faults is presented.Experimental studies show that the proposed method can wear fault and broken tooth fault classify and gear fault classification method to contrast the empirical mode decomposition and support vector machine,fault recognition rate increased significantly,the accuracy reached almost 100%.(3)Proposed a method for gear fault diagnosis,and tracking based on local mean,create a Gabor dictionary,sparse signal representation of,combined with the spectral kurtosis principle to select the optimal component,spectral analysis,the results show that this method can shorten the signal reconstruction time,speed up the running speed,the accurate extraction of the characteristic frequency of fault.(4)Proposed a method for gear fault diagnosis and tracking of local mean decomposition orthogonal matching based on graph effect contrast before and after the signal reconstruction is seen,the number of fault feature extraction method in this paper is obviously increased,in the process of signal reconstruction in removing the interference information,significantly improve the accuracy of diagnosis.