Researches On Bearing Fault Diagnosis Of Permanent Magnet Synchronous Motor Under Non-stationary Conditions

Motor bearing fault is one of the most frequent faults in Permanent Magnet Synchronous Motor(PMSM).When small cracks appears on the raceway and rolling element of the motor bearing due to the inherent internal stress or pollution,the alternating stress of the bearing during operation will make the cracks gradually expand,resulting in noise and torque fluctuation,and even make the bearing fail in serious cases.Vibration signal or current signal is the most commonly used signal in bearing fault diagnosis.However,the need to install additional sensors and low diagnostic accuracy limit the application of vibration signal and current signal in fault diagnosis.When bearings in electromechanical equipment run under steady state condition(constant speed,constant load condition),there is a constant bearing fault characteristic frequency spectrum peak in the signal frequency domain.However,in the industrial field,the bearing often operates in the unsteady condition of changing speed and torque,the periodic characteristics of the signal are destroyed by the time-varying condition,and the frequency components in the spectrum are constantly changing,confusing each other,which greatly increases the difficulty of fault diagnosis.In this paper,based on the analysis of the existing bearing diagnosis technology,the PMSM bearing fault diagnosis method based on observer estimated signal and high frequency resonance signal of rotating speed under non-stationary conditions is deeply studied.The main research content of this paper is as follows.Firstly,the bearing fault geometry model is established according to the bearing geometric structure and motion law,and the mechanism of the disturbance component of the fault characteristic in the motor torque when the motor bearing fault occurs is analyzed according to the motor torque equation and the closed-loop control algorithm.Then,the influence of the closedloop control system on the amplitude of the bearing characteristic component in the motor speed signal is revealed by establishing the small signal model.In order to solve the problem that the frequency component of fault signals changes with time under unsteady working conditions,it is difficult to use traditional signal processing algorithms for diagnosis.An angular resampler algorithm is introduced to pre-process the Angle and speed signals,and convert the unsteady temporal signals into steady angular signals for fault diagnosis.Secondly,according to the bearing fault characteristic signal(BFCF)is contained in the motor speed signal,an extended-electromotive force(EEMF)based speed sensorless observer algorithm is proposed to diagnose the bearing outer raceway fault.Based on the voltage and current signals of the PMSM combined with the motor model,an observer for estimating the rotational speed of the motor was constructed.The angular resampling of the rotational speed signal was performed using the rotational angle information and the order spectrum was calculated.The fault threshold generation algorithm based on the value of the spectrum was used to interpret the fault existence in the order spectrum.The experiment verifies that the estimated speed signal component can be used for the diagnosis of the bearing outer raceway fault under the condition of varying speed and torque.The comparison experiment shows that the diagnosis result using estimated speed signal component is better than that of traditional current and voltage signal.Thirdly,by analyzing the transmission path of the motor torque signal,the transmission path of the torque signal is shorter than that of the speed signal,and the interference is lower.On this basis,a fault diagnosis method of bearing outer ring based on natural torque observer is proposed.The q-axis current multistage difference algorithm is used to suppress the jitter of estimated value under stationary conditions and accelerate the convergence under non-stationary conditions.The load torque signal reconstructed from the voltage and current signal is calculated to generate the order spectrum after amplitude modulation compensation and angular resampling.When determining whether the peak amplitude of the order spectrum exceeds the standard,to further reduce the influence of human factors,the optimal window length is calculated iteratively by information entropy.The method is verified by simulations and experiments under the condition of variable speed and variable torque,and compared with phase current and q-axis current signal.The results show that the effectiveness and SNR of the proposed method are superior to the traditional motor bearing failure indicators.Finally,in view of the difficulty of detecting the fault characteristic signal under alternating torque modulation when the bearing inner raceway fault occurs,the double-inertia system of the motor is modeled,and the mechanism of the high-frequency resonance component of the speed signal generated by the double-inertia system under the excitation of pulsed torque disturbance is analyzed,as well as the modulation effect between the resonance component and the bearing fault characteristic component is analyzed.A fault diagnosis method of bearing inner and outer raceway based on the resonance component of rotational speed signal is proposed.In this method,the artificial bee colony algorithm is implemented for the parameter optimization of the variational mode decomposition.When the components unrelated to the fault signal are filtered out,the multipoint optimal minimum entropy deconvolution adjustment algorithm is then introduced to enhance the order spectrum amplitude corresponding to the bearing fault.The simulation and experimental results show that the bearing outer and inner raceway faults can be diagnosed by the proposed method under non-stationary conditions,and the comparison experiments with other methods also proved that the proposed method has the advantage of higher signal-to-noise ratio of diagnosis results.