Gear Fault Diagnosis Research Based On Theory Of EMD And Morphology

Taking the gear as research object,the paper focuses on the problem of end effect in EMD algorithm,gear vibration signal characteristics and the gear fault signal detection issue under strong noise with the method of empirical mode decomposition and morphological filtering.Firstly,the article gives a gear fault simulation mode and commonly used gear vibration signal analysis method based on the vibration mechanism.Then the paper focuses on the signal processing method on inhibiting background noise and highlighting fault characteristics aiming at the problem of the fault signal detection issue under strong noise.Secondly,the sifting process of the EMD method and the cause of the endpoint effect is described detailedly,and the defects of the existing endpoint effect suppression method are analyzed.Then,aiming at the problem of end effect in EMD algorithm,a novel method that Support Vector Machine(SVM)extension and window function combined effects is proposed on the basis of studies and summarizes old methods.This method makes up for the SVM continuation still can't find the endpoint and window function will change the original signal.The algorithm is simple,fast,and can effectively solve the problem of endpoint effect.Thirdly,the basic principle and function of morphological filter is described,and aiming at the problem of selecting the structural elements of morphology,a method of basing on kurtosis criterion to optimize structural elements of the adaptive algorithm has been proposed.The simulation signal and the result of gear fault experiment show that the method has the obvious effect for signal de-noising processing and impact feature extraction.At last,EMD algorithm and morphological filter which is used in gear fault feature extraction is researched.Aiming at the defects in the existing EMD noise reduction algorithm,based on the difference of the noise signal and the general signal in the autocorrelation function,a noise reduction algorithm based on EMD modalcorrelation and morphological filtering is proposed to construct a new criteria for judging demarcation points.In order to keep the high frequency components of the useful signal in the noise component as much as possible,the noise pattern is determined by morphological filtering.The simulation signal and the results of gear fault experiments show that the method can better distinguish the cutoff point of modal noise and useful signal.The method not only can highlight the gear fault characteristics more effectively but also have a significant effect in signal de-noising through comparing with the method of adaptive morphological.To sum up,this paper puts forward a kind of improved algorithm to restrain endpoint effect innovatively,providing a train of thought and method for further research work of restraining endpoint effect.Introducing the adaptive morphological filtering to the gear fault diagnosis research and combining with the EMD de-noising algorithm and providing a new train of thought for gear fault diagnosis.This method has great practical value and significance.