Drive Induction Motor Noise Source Analysis And Identification Method Research

The main noise source of the drive system is the induction motor,and it is crucial to systematically analyze the various types of noise and separate the sources to facilitate induction motor condition detection,noise reduction treatment,and fault diagnosis.The noise sources of induction motors overlap to some extent and radiate outward through the housing,making it extremely challenging to differentiate between these sources.The characteristics of induction motor noise sources and the methods for identifying induction motor noise have practical value in controlling motor noise and play an essential role in advancing research on engineering acoustics.This thesis conducts a comprehensive analysis of the mechanism behind different types of vibration noise in induction motors,focusing on the air gap harmonic magnetic field and its influencing factors.The effects of various harmonic magnetic fields on the radial electromagnetic force are also examined.Subsequently,the thesis performs analytical calculations to determine the induction motor’s electromagnetic,mechanical,and aerodynamic noise characteristics.Additionally,a simulation is used to construct an electromagnetic model of the induction motor,validating the accuracy and reliability of the analytical model.The simulation results indicate that the tangential effect significantly influences electromagnetic noise during induction motor startup and strong armature response.The empirical wavelet transform and variational modal decomposition methods based on classical scale space were found to be insufficient in constructing inputs for single-channel blind source separation.Therefore,this thesis proposed a novel method based on adaptive scale-space mode extraction for separating and identifying single-channel induction motor noise sources.A reasonable a priori assumption was used to transform the frequency domain signal into a Markovian transition field,and an electromagnetic characteristic scale space was constructed by incorporating electromagnetic noise characteristics.This allowed for determining the optimal number of signal modes and predicted center frequency.Adaptive penalty factors were then formulated using aggregation factors to accurately extract harmonics and shock modes present in the induction motor noise signal.Simulation signals were used to validate the proposed method,and the results demonstrated its ability to overcome the issues of over-decomposition in classical scale space variational modal decomposition and the challenge of balancing harmonic and impact modes.Furthermore,the adaptive scale-space mode extraction method was employed to construct a multi-channel input for blind source separation using induction motor noise tests.The validity of the modal components was assessed through correlation and variance contribution rate analysis.Robust independent component analysis was applied to extract independent noise components,which were subsequently identified using power spectral density and envelope analysis.Finally,based on the basic principles of blind source separation,the contributions of different independent noise sources were calculated using the obtained unmixing matrix.Experimental results confirmed that the proposed method accurately and effectively separates induction motor noise sources and successfully identifies electromagnetic noise of different orders,aerodynamic noise,and switching frequency noise.