Research On Gearbox Fault Diagnosis Based On Blind Source Separation

In this paper,the blind source separation algorithm and signal denoising method are studied for the problem that the gearbox composite fault is difficult to extract.The singular value denoising algorithm and the EEMD-SVD-JADE single-channel blind source separation algorithm and its application in the experiment are analyzed.Firstly,the gearbox vibration model and blind source separation theory are introduced.The failure modes and vibration characteristics of the gears,the fault characteristics of the bearings,and the vibration model of the gearbox were studied.The commonly used time domain,frequency domain and time-frequency analysis methods for fault diagnosis of gearboxes are summarized.Explained the blind source separation mathematical model;the characteristics of blind source separation and pretreatment methods.Focusing on FastICA algorithm and JADE algorithm,the evaluation index of separation effect is given.Simulations prove that these two algorithms are effective in the complex fault diagnosis of gearboxes.Secondly,the single channel blind source separation algorithm based on EEMD-SVD-JADE in the virtual channel method is discussed.The single-channel signal is converted into multi-channel signals by EEMD decomposition.The number of source signals is estimated based on the SVD source number estimation method.And the effective IMF component reconstruction signal is selected according to the correlation coefficient,kurtosis and variance criteria.Blind source separation of reconstructed signals use JADE algorithm The simulation of the composite fault signal of the gearbox verifies that the algorithm can obtain good separation effect.Finally,the blind source separation problem in noise environment is studied and the algorithm is applied to the experiment.Random interference noise will reduce the effect of blind source separation.By comparing the three methods of autocorrelation noise reduction,singular value difference spectrum noise reduction and improved singular value difference spectrum noise reduction,the blind source separation after the improved singular value differential spectrum denoising is best.The gearboxcomposite test fault signal was measured with a single sensor on the gearbox test bench.The composite fault feature of the measured gearbox was extracted by singular value noise reduction and EEMD-SVD-JADE blind source separation algorithm.In summary,the singular value noise reduction and EEMD-SVD-JADE single-channel blind source separation algorithm is applied to the composite gearbox fault diagnosis,which provides a new idea for gearbox fault diagnosis.