Load Identification Methods Of Rotor System Based On Vibration And Current Signal

As the most common mechanical equipment,rotating machinery plays a vital role in every manufacturing industry.The rotor system is the core component of the rotating machinery.Its stable and normal operation is related to the safe production of the rotating machinery and even the entire mechanical equipment,it is also related to the operators' safety.In rotating machinery,when the load of the rotor system exceeds its maximum load bearing value,serious consequences may occur.Therefore,it is especially important to identify the load of the rotor system.The rotor system generates vibration,and the stator current in the motor that supplies power changes correspondingly under load.Therefore,in order to realize rotor system diagnosis and online monitoring,it is significant and practical to study the qualitative and quantitative identification of rotor system load based on vibration signal and motor current signal.Firstly,by simplifying and discretizing the rotor system,the turntable and bearing are subjected to force analysis and mechanical energy is calculated,and then the Lagrange equation is used to solve the vibration differential equation,and the bending-torsional coupling model is established.At the same time,the rotor system and the motor are connected to establish an electromechanical coupling model.Based on the coupled model,a simulation model is built in the MATLAB/Simulink environment.Five loads(four types of superposed loads and random load)are applied to the rotor system through Signal Builder to obtain the turntable vibration signal and the motor current signal.By analyzing time domain of two types of signals,it is found that the vibration signal and the motor current signal change are consistent with the load change.Secondly,the rotor system load is qualitatively and quantitatively identified based on the vibration signal.In the process of qualitative recognition,Fast Fourier Transform(FFT),Singular Value Decomposition(SVD)and reconstruction,Ensemble Empirical Mode Decomposition(EEMD),Intrinsic Mode Function(IMF)energy percentage calculation and Probabilistic Neural Network(PNN)methods are proposed to realize the qualitative identification of rotor system load.In the process of quantitative identification,the Extreme Learning Machine Regression(ELMR)method is adopted.According to the different eigenvalues of vibration signals,different load sizes are regressed to realize the quantitative identification of rotor system load.At the same time,the rotor system load is qualitatively and quantitatively identified based on the motor current signal.In the process of qualitative identification,FFT,SVD and reconstruction,Wavelet Packet Decomposition,node energy percentage calculation and Radial Basis Function(RBF)Neural Network are proposed to realize the qualitative identification of rotor system load.In the process of quantitative identification,a BP Neural Network optimized by Genetic Algorithm(GA-BP)is proposed.According to different motor current signal eigenvalues,different load sizes are regressed to realize the quantitative identification of rotor system load.Thirdly,the rotor system is quantitatively identified based on the multi-source information fusion method of the vibration signal and the motor current signal.Using the single source signal,the relationship between the steady load and the eigenvalue(of the vibration signal and the motor current signal)is taken as the training sample.The Support Vector Machine Regression(SVMR)model is established after optimizing the parameters,and different load sizes are regressed according to different eigenvalues.Then multi-source information fusion is performed to quantitatively identify the load.Finally,the rotor system test bench is built,and the test scheme is developed and implemented to obtain the rotor system turntable vibration signal and the motor stator current signal under different loads.Based on the vibration signal,motor current signal and information fusion methods,the rotor system loads are identified.The simulation results and test results are compared and analyzed.The results verify the feasibility of the simulation research process and the correctness of the load identification methods,which provide theoretical and experimental support for load identification and online monitoring of rotating machinery.