Research On Compound Fault Feature Identification And Compression Diagnosis Of Vibration Signals

Compound faults of rotating machinery are not obvious because of their different faults,coupling between faults and noise,etc.At the same time,intelligent diagnosis based on big data cannot take into account the problems of diagnosis accuracy and efficiency.Therefore,this paper takes the composite fault vibration signals of key parts of rotating machinery(rolling bearings and gears)as the research object,and studies the problems of complex fault feature recognition and the shortcomings of intelligent diagnosis network from the two perspectives of fault feature recognition of signal processing and intelligent fault diagnosis method based on big data.The research content of this paper mainly includes the following aspects:(1)In order to solve the problems of mutual coupling between composite faults of rotating machinery,weak faults in composite faults are not easy to be identified,dominant frequency of fault features is not obvious,and noise makes interference frequency too much,a new fault feature recognition method of rotating machinery based on blind source separation and improved simplified convolution sparse filter FIVA-CSF is proposed in this paper.In order to verify the performance of this method,the composite fault signals of rolling bearing and gear collected in the laboratory are compared with a variety of existing signal processing methods.(2)In order to solve the problem that the traditional blind source separation method can not adapt to the separation of composite fault signals under the condition of underdetermined,and weak faults are easy to be affected and difficult to be extracted in the process of signal upgrading.Combined with the weak fault impulse enhancement method,this paper proposes a single channel feature recognition method MEFC,and analyzes the composite fault signals collected by single channel to verify the advantages of this method in the case of single channel feature recognition.(3)Aiming at the problems of large amount of data and low diagnosis efficiency in the deep neural network diagnosis process under the background of big data,this paper proposes a fast compression diagnosis method based on compressed sensing technology and convolutional neural network,and applies it to the recognition of composite fault.Firstly,it is proved that the compressed signal contains the fault information needed in the diagnosis process.By analyzing the similarity between the reconstructed signal and the original signal and the degree of fault information retention,the validity of the compressed signal is verified under a certain compression ratio.Then,the various fault signals(single and compound faults of rolling bearing and gear)collected in the laboratory are compressed and observed,and the signals under different compression ratios are obtained.Finally,through the experiments of various signals under different compression ratio,it is verified that the compression diagnosis method in this paper can give consideration to the accuracy and efficiency of diagnosis to a certain extent.