Research On Bearing Fault Diagnosis Method Based On Hilbert-Huang Transform

At present,the industrialization development of our country is still in the middle and later stage.The technology of fault diagnosis is of great significance for promoting the industrialization process of our country.According to statistics,the damage of rolling bearing is one of the main factors that lead to mechanical equipment failure,and the effective extraction of fault characteristics of rolling bearings plays an important role in accurate diagnosis of bearing fault diagnosis.However,in the actual environment,due to the influence of surrounding complex environment,the fault signal collected is often characterized by strong noise,weak fault signal and difficult fault feature extraction.Therefore,the research and analysis of the fault feature extraction of rolling bearing is one of the main parts of the fault diagnosis of the whole rolling bearing.In order to solve this problem,this paper uses the advantage of the algorithm in the analysis of such fault signals,considers the effect of the algorithm defect on the fault feature extraction,and improves the fault feature effectively from the following aspects.The feasibility and effectiveness of the improved algorithm is verified by experimental simulation.The main research work in this paper is as follows:First of all,this paper uses the Hilbert yellow transform(HHT)algorithm as the research foundation to understand the structure of the HHT algorithm.The mechanism and function of the components are deeply studied and expounded,and the advantages and disadvantages of the algorithm are summarized and analyzed in detail.Aiming at the effect of signal endpoint effect on fault feature extraction after EMD decomposition,this paper proposes an endpoint effect suppression method based on ELM mirror closed continuation.An extreme learning machine prediction(ELM)model is established to predict the extremum points at both ends of the signal through the continuous learning and training of the signals,so as to achieve thepurpose of extending the original data sequence.The mirror close extension method makes the new data sequence form a closed loop signal from the end to the end,and eliminates the data endpoint from the "interior" to suppress the endpoint effect.Through the experiment simulation,and to illustrate the applicability of the algorithm to a variety of signals,the analysis of different fault signals,and verify that the proposed algorithm has an obvious effect on the suppression of endpoint effect.Secondly,due to the algorithmic mechanism or EMD decomposition,complex signals will produce false components which can't represent fault signal characteristics or exist in original signals.These false components(IMF)will disturb the effective extraction of fault features and affect the results of fault diagnosis.Therefore,this paper presents the research content of EMD false components accurately,and puts forward the K-L divergence method and applies it to the practical problem of fault feature extraction.Simulation results show that the proposed algorithm has higher accuracy than classical correlation coefficient method,and is helpful for effective extraction of fault features.Finally,in order to realize the accurate identification and diagnosis of the fault degree,this paper puts forward a bearing fault diagnosis method based on multi feature quantity.First,a set of IMF components is obtained by using empirical mode decomposition(EMD)for the fault signal.Each component is calculated according to the kurtosis and correlation coefficient criterion.A set of Kr values is obtained through the corresponding relation between the two parameters,and a group of IMF components strongly related to the original signal are selected.Secondly,it reconstructs the component and extracts and analyses the multi feature quantity of the time domain parameters,the singular value and the energy entropy of the AR model parameter matrix,constructs the multi feature parameter matrix and input the support vector machine to establish the multi classification prediction model.The simulation results show that the method can accurately identify and diagnose the fault level andis an effective method for rolling bearing fault diagnosis.