| Bearing is the key part of mechanical equipment,whose state directly determines whether the mechanical equipment can operate normally.It is of great significance to find an effective bearing fault diagnosis method to ensure the healthy operation of mechanical equipment.In this paper,aiming at how to diagnose the fault of rolling bearing accurately,a sparse representation based fault diagnosis algorithm of rolling bearing is proposed.The application of Maximum correlation kurtosis deconvolution algorithm and K-Singular value decomposition algorithm in compression sensing method,sparse representation based fault diagnosis method of rolling bearing and its application in experiment are mainly studied.First of all,this paper introduces the causes of rolling bearing faults and the theoretical characteristic frequencies of different fault types in detail,and summarizes the commonly used time-domain analysis,frequency-domain analysis and timefrequency analysis methods in vibration signal analysis and processing.Secondly,the sensing method of vibration signal compression based on MCKD is studied.Based on the compressed sensing algorithm,this method uses the MCKD algorithm to preprocess the signal in order to improve the sparsity of the signal.Simulation and experiment show that this method can effectively improve the signal sparsity.Thirdly,the vibration signal compressed sensing method based on K-SVD dictionary learning is studied.Two kinds of dictionaries based on K-SVD training and fixed Dictionary of DCT are used in signal reconstruction,and the results are compared and analyzed.The simulation and experimental results show that the original dictionary generated from the original signal can effectively improve the similarity between the reconstructed signal and the original signal after K-SVD training.Finally,the sparse representation algorithm is applied to the fault diagnosis of rolling bearings.The sensitivity of feature indexes to different fault features is different,and the quantized value of insensitive fault features is zero or close to zero.The process of quantization is similar to the process of sparse representation.Therefore,feature indexes and sparse representations can be combined,and dictionaries of different fault states can be generated using the calculated feature indexes.The generated different fault state dictionary is used to reconstruct the signal.When the state dictionary matches the state of the reconstructed signal,the reconstruction error is extremely small,and vice versa.According to the reconstruction error,the fault type of rolling bearing can be determined,and the experiment shows that the method can effectively judge the state of rolling bearing.In this paper,sparse representation method is applied to the fault diagnosis of rolling bearing,which provides a new idea for the fault diagnosis of rolling bearing. |
