Research On Fault Diagnosis Of Bearing Based On Vibration Signal Denoise And Decomposition

Rolling bearing is a kind of parts,it exists in all kinds of mechanical equipment.It is also very prone to failure parts.The defectiveness of rolling bearings can result in abnormal vibration and noise of equipment,even seriously damage to the equipment and catastrophic accident.We think the research on fault diagnosis has great meaning both in theory and reality.This paper summarizes procreant cause of faults of rolling bearings in theory,constructs models of vary fault states.Some of the common bearing fault diagnosis methods and evaluation indexes are introduced.Through the research literature at domestic and abroad,a kind of bearing fault diagnosis based on singular value decomposition(SVD),local characteristic scale decomposition,fuzzy entropy and the hidden markov model is proposed.First of all,due to the bearing fault vibration signal contains a large number of random noise,it is necessary to denoise the signal.In this paper,by studying the theory of singular value decomposition,a new effective method for determining effective rank order of singular value decomposition denoising based on k-means clustering through the singular value decomposition to remove random noise in the signal components and to prepare for subsequent fault diagnosis.Secondly,to solve the problem in bearing fault of nonlinear and non-stationary and obtaining bearing fault characteristics.A self-adaptive method,the local characteristics of scale decomposition is proposed.Aiming at the mode mixing problem in local characteristics of scale decomposition,by comparing the extremum continuation,polynomial continuation,appropriate methods can be got to restraint extremum.Through the study of the structure of the baseline signal LCD method was improved and improve the accuracy of the decomposition of the method.Thirdly,in order to determine the fault types of rolling bearings,the characteristic parameters of rolling bearing fault and the appropriate training learning model are required.In this paper,the fuzzy entropy is used as the characteristic parameter,and the fuzzy entropy of the ISC of the vibration signal after LCD is obtained,and then it is put into the hidden Markov model for training and learning to determine the fault type of the rolling bearing.The feasibility of this method is proved by the experiment.This paper studies the signal noise reduction and fault feature extraction to provide a valuable reference for bearing fault diagnosis in gearboxes.