Research On Rotor Crack Fault Diagnosis Based On Variational Mode Decomposition

Rotating machinery is the most widely used kind of mechanical equipment in modern industry.The rotor,the core component of rotating machinery,often produces crack faults in various working conditions.There have been many catastrophic accidents caused by rotor crack faults at home and abroad,resulting in huge economic losses.Therefore,it is of great significance to research diagnosis of cracked rotors.In this paper,the vibration displacement signals of cracked rotor are not stable and nonlinear,the fault characteristics are difficult to extract and the collection and arrangement of the samples is lack in the early period of failure,so a method based on the combination of Variational Mode Decomposition and Auto Regressive model is proposed to diagnose of cracked rotors.The main contents of this article are as follows:1.The crack is simulated by laser cutting transverse cracks on the rotor.The vibration test of cracked rotor is carried out by using the Donghua DHRMT rotor test rig.By comparing and analyzing the vibration signals of the normal rotor,it is observed that harmonic component of 1×,2×,3× et al.The vibration signals of the normal rotor show 1× and a smaller amplitude 2×.The reason is that there is a certain bending deformation in the rotor,but it is within the allowable error range of the experiment.2.The basic principle of VMD is expounded.Through the simulation and analysis of multimodal signal,shock signal,discontinuous signal and AM-FM signal,it is found that the VMD is superior to EMD in the anti impact and anti multimodal aliasing.The experiment further verifies the advantage of VMD in dealing with the vibration displacement signal of cracked rotor.3.A fault diagnosis method for rotor crack based on VMD and AR model is proposed.VMD is used as a preprocessing process for establishing AR model.After obtaining the model,the model parameters and the variance of the residual difference are determined as the state eigenvectors,and then the cracked fault diagnosis is compared by comparing the Mahalanobis distance between the eigenvectors.Finally,the feasibility of the method based on VMD and AR model is verified by analyzing the experimental data.