Fault Diagnosis Of Gearbox Based On MED-EMD And The Diagonal Slice Of Bispectrum

Gear box is a major part of the transmission power in rotating machinery.Such as shaft,bearing and gear,the rotating parts of gear box often run in a environment which has a complex variable load.Because of fatigue,these parts damage resulting in weak fault,but under strong background noise,the fault signal is often difficult to detect.It would caused serious consequences if the weak fault had developed to serious fault.Therefore,choosing the appropriate method to reduce noise will has a great significance for early fault feature extraction.For mechanical signal processing,the non-Gaussian and non-stationary vibration signals were ignored,which come from mechanical equipment failure,and the traditional fault diagnosis methods can’t suppress Gaussian noise on the influence of the fault signal,which is often not to extract non-stationary and non-Gaussian characteristics of weak fault characteristics at the beginning of the fault.In view of the early fault signal of gear box having the characteristics of nonlinear,non-stationary,non-Gaussian and vulnerable to strong background noise,the combining method to extract the weak fault feature is proposed in this paper,which based on minimum entropy deconvolution(MED),empirical mode decomposition(EMD)and the diagonal sliced bispectrum.The main research content is as follows:1)Introduced the principle of MED,EMD,mutual information and the diagonal slice of bispectrum and their connection to others.On the basis of the introduction,the simulation signal is analyzed by the method of MED-EMD and the diagonal slice of bispectrum.The first three IMF components from MED-EMD is analyzed by the diagonal slice of bispectrum,which found that the carrier frequency 1)1= 70 Hz appeared in the low frequency,and the modulation frequency 1)2= 200 Hz with the frequency of clusters,the double frequency of 1)2with its frequency clusters and carrier frequency 1)9)= 300 Hz with its frequency cluster were marked.Comparing with other two methods,MED-EMD and the diagonal slice of bispectrum is the best of them to reduce noise,which verified the method based on MED-EMD and the diagonal slice of bispectrum for the rotating equipment fault diagnosis is feasible.2)In this paper,through MED-EMD to reduce noise,original signal is decomposed into several intrinsic mode functions(IMF).MED as the outer-loop of EMD can compensate for the shortcomings of the EMD decomposition under strong background noise,the IMF components having strong correlation with original signal are selected as effective components and analyzed with diagonal slice of bispectrum to extract pitting fault feature.The diagonal slice of bispectrum analysis can suppress the interference of Gaussian noise on the IMF components.At the same time,comparing with t he results between this method and the conventional power spectrum method based on EMD verifies the practicability and superiority of this method.3)The simulation signal,having characteristics of impact and modulation,has been designed to improve the algorithm;Structures gearbox test rig,collect the vibration signal that contains information in gear pitting and use the method to analyze it.The result verifies the method based on MED-EMD and the diagonal slice of bispectrum is practical when applied it to fault diagnosis of rotating equipment;Test to the fault signal of a certain type of wind gearbox with actual operation,extract the weak fault feature,make the corresponding fault diagnosis,match the gearbox check results and The results verify the judgment that the fault is bearing inner ring pitting in the intermediate gear shaft on the motor side.At the same time,comparing with the traditional power spectrum method based on EMD,it shows that this method has good applicability to extract the weak fault feature.