Research On Fault Diagnosis Of Loss Of Excitation Fault Of Permanent Magnet Synchronous Motor Based On Information Fusion

In order to use water resources more effectively,water-saving irrigation technology is widely used in agricultural production,and water pump motors play a pivotal role in water-saving irrigation systems.Due to the poor working conditions in the agricultural irrigation process,the motor is prone to the demagnetization fault.If it is not handled in time,the motor performance w ill be reduced.Seriousl y,safety accidents may be involved.At present,most of the current diagnosis methods of the demagnetization fault are based on the characteristic information of a single signal.The fault information cannot be characterized from m ultiple angles,and it i s difficult to accurately diagnose the demagnetization fault,that causes misdiagnosis.Based on the above problems,this topic takes permanent magnet synchronous motors commonly used in agriculture as the research object.The torqu e signal generated by the tangential electromagnetic force and the vibration signal generated by the radial electromagnetic force have complementary characteristics.The characteristic information is fused to realize the accurate diagnosis of the local demagnetization fault of th e motor.The main contents of the thesis research are as follows:(1)Under the demagnetization fault,the tangential electromagnetic force and the radial electromagnetic force,causing the changes of the torque and vibration signal,are theoretically deri ved.Then the normal operation of the motor and the demagnetization fault of a single permanent magnet under load conditions are carried out based on Maxwell and Workbench software.The faults with 30%,50%,70%,and100% degree of the torque and vibration are respectively simulated and analyzed,and the regulation of their changes with the degree of the demagnetization fault is analyzed,and it is verified that the demagnetization fault will lead to changes in the torque signal and vibration signal.(2)Based on theoretical and experimental requirements analysis,the experimental motor is simulated for the demagnetization fault,and the YH502 torque sensor and MPU-6050 acceleration sensor are selected to build a data acquisition syst em to achieve real-time acquisition and analysis of torque and vibration signals.(3)In view of the noise interference problem of the collected signal,the wavelet threshold denoising method is used to preprocess the torque and vibration signals under the normal operation of the motor and the single permanent magnet demagnetization faults of 30%,50%,70% and 100%.Then the characteristic information in the time domain and frequency domain is extracted and analyzed from the signals.It is found that the re sults are basically cons istent with the simulation data,which verifies the feasibility of the experimental protocol.(4)The RBF neural network model for the diagnosis of the demagnetization fault is constructed.The information fusion of torque signal an d vibration signal is us ed to realize the prediction of local demagnetization fault,and compared with the traditional single signal.The experimental results show that under the same sample conditions,the relative error of the information fusion of the torque signal and the vi bration signal on the permanent magnet demagnetization fault prediction is only 3%.Compared with that of the traditional single vibration signal and single torque signal,the relative error is reduced by 8% and 15%,respectively.I t is verified that the i nformation fusion method of torque signal and vibration signal can improve the accuracy of motor demagnetization fault diagnosis.