Research On Fault Diagnosis Method Of Rotor System Based On Time Varying Filtering Based Empirical Mode Decomposition

Turbine,compressor,gas turbine and other rotating machinery are common types of equipment in industrial production,which have important applications in the industrial field.Research on effective methods to monitor the health status of equipment and diagnose the type,location and severity of faults is of great significance for ensuring the normal operation of machinery and equipment,avoiding economic losses and casualties.Vibration diagnosis method is widely used in the field of rotor system fault diagnosis,which mainly includes fault mechanism,fault feature extraction and fault pattern recognition.Therefore,on the basis of time varying filtering based empirical mode decomposition,the fault diagnosis of rotor system is studied from the above three aspects.Firstly,the fault characteristics of common rubbing,cracking,oil whirl and unbalance faults of rotor system are summarized.The dynamic model of rotor system under typical fault state is established.The differential equations of motion of each fault are solved by computer numerical analysis,and the simulation signals of each fault are obtained,thus laying a theoretical foundation for the research of rotor fault diagnosis.In order to effectively extract the characteristics of rotor fault signals,the time varying filtering based empirical mode decomposition(TVFEMD)method is applied to fault diagnosis of rotor systems.Firstly,aiming at the blindness of parameter selection in this method,the parameter optimized TVFEMD method is studied.Then,the parameter optimized TVFEMD and Hilbert transform are combined as a time-frequency analysis method to extract fault features from rotor fault signals.Finally,the analysis of simulation signal and experimental fault signal shows that the time-frequency analysis method can effectively extract fault features from fault signals,and has strong advantages compared with existing methods.Based on the time-frequency analysis method of the parameter optimized TVFEMD combined with Hilbert transform,according to the difference of time-frequency distribution characteristics of different fault signals,a method of fault type recognition based on information entropy and particle swarm optimization support vector machine(PSO-SVM)is studied.Firstly,the parameter optimized TVFEMD and Hilbert transform are used to obtain the Hilbert time-frequency diagram of the rotor fault signal.Then,the partial band energy entropy of the time-frequency diagram is calculated.Finally,the entropy value is used as the eigenvector to input the support vector machine(SVM)to recognize the rotor fault.The diagnostic results of simulation and measured rotor fault signals show that this method can accurately distinguish typical faults of normal,unbalanced,rubbing,oil whirl and other rotor systems,and is not affected by rotor speed.Compared with EMD combined with time-frequency entropy method,this method has higher diagnostic accuracy.