Research On Fault Feature Extraction Method Of Rotating Machinery Based On Blind Separation Technology

The intelligence and complexity of rotating machinery poses a huge challenge to fault diagnosis technology.As the core of fault diagnosis technology,fault extraction technology directly affects the fault diagnosis process and even the performance of the entire diagnostic system.Therefore,the research of fault feature extraction technology is of great significance.Aiming at the problem of non-stationary vibration signal of rotating machinery and difficult extraction of weak fault signal features,and solving the generality of diagnostic system,in this paper,blind source separation(BSS)and blind deconvolution(BD)techniques are applied to the frequency domain fault feature extraction of single-channel signal and multi-channel signal of rotating machinery.Furthermore,a general fault feature extraction method for rotating machinery is proposed.The effectiveness of this method in feature extraction of weak fault signals and non-stationary fault signals is verified by analyzing various fault modes of gears,rolling bearings and rotors.The specific research contents are as follows:(1)The theory of blind source separation and blind deconvolution and the preprocessing method before blind source separation is introduced.Combined with the algorithm of classical blind separation technique,the performance evaluation index of blind source separation algorithm is studied,and a new frequency domain performance evaluation index is proposed.(2)The frequency domain fault feature extraction method for single channel signals is studied and a single channel ESK method is proposed.Through the faulty gear experiment,based on the fault mechanism of the gear and the analysis of the frequency domain features of the signal corresponding to the common fault,the effect of the single-channel ESK method proposed in the frequency domain feature extraction of the fault signal is verified.(3)Through the fault rolling bearing experiment,based on the fault mechanism of the rolling bearing and the frequency domain characteristics analysis of the signal corresponding to common faults,the feature extraction effect of the single-channel ESK method proposed in this paper is characterized by the non-stationary characteristics of the rotating machinery vibration signal.(4)The fault feature extraction technology of rotating machinery vibration signal under the condition of rolling bearing fault and gear fault is studied.The feature extraction effect of the single-channel ESK method proposed in this paper on multi-component faults in complex rotating machinery is verified.(5)Based on the rotor fault test platform,based on the rotor fault mechanism and the frequency domain characteristics analysis of the signal corresponding to common faults,the feature extraction effect of the single-channel ESK method proposed in this paper is valid under the condition of low signal-to-noise ratio(SNR)of the rotating machinery vibration signal.(6)The multi-channel signal fault feature extraction method for large rotating machinery is studied and a multi-channel SBD method is proposed.Through the multi-channel rotor failure experiment and multi-channel rolling bearing failure experiment,the feature extraction effects of the single-channel ESK method and multi-channel SBD method proposed in this paper are compared and analyzed.(7)Based on the Lab VIEW platform,the fault feature extraction system of the single-channel ESK method and the multi-channel SBD method proposed in this paper are implemented,which is convenient for the application of the feature extraction method proposed in this paper.