Research On Fault Diagnosis Method Of High Voltage Isolating Switch Based On Multi-channel Vibration Signal Fusion

The isolating switch not only ensures normal use of electricity,but also isolates the fault point when the system is faulty,so it has important practical value for fault monitoring.In this thesis,a three-dimensional model of GW4B-252DW type high voltage isolating switch was established by SolidWorks,which was imported into ANSYS Workbench for finite element modal analysis,and the natural frequency and mode of the isolating switch were calculated under three states:closing,opening and inadequate closing.The dynamic model of transmission mechanism of the high voltage isolating switch was established by virtual prototype simulation software ADAMS,and the dynamic process was simulated for opening and closing according to the actual working conditions.An exploration was made in the variation trend of mechanical characteristics under common faults,such as loosening and jamming of connecting rod of switch.The variation characteristics of driving moment waveform and contact force were obtained under different faults.A fault diagnosis method of high voltage isolating switch based on vibration signal was also proposed.A data extraction device was developed for the high voltage isolating switch,and voltage and current signals of vibrating motor and conducting motor were collected by the sensors.The online monitoring system based on LabVIEW platform was developed to realize the functions of signal extraction,storage,processing and display.Fault simulation test was carried out on site for GW4B-252DW isolating switch in seven fault conditions,i.e.high operation voltage,low operation voltage,loose transmission device(crank arm,inter-electrode and inter-phase),loose bolt and loose locking.6 vibration signals and the voltage and current signals of the conducting motor were collected under normal state and simulated fault,and 8 types of typical state databases were established.In light of the vibration signals of isolating switch,the method of combining time domain and frequency domain is used to extract characteristic factors.The vibration signals were preprocessed separately,the feature factor was extracted and merged,the factor weight was calculated and filtered,and the fault signature database was established;morphology and wavelet packet method are used to de-noising the vibration signal,and the vibration amplitude maximum point and the empirical mode decomposition energy moment are extracted as the time-frequency characteristic factors of the vibration signal.Due to the fact that the characteristic factor of single vibration signal is difficult to identify the loosening fault of transmission mechanism,a feature factor extraction method is proposed to merge the characteristics of multi-channel vibration signals.The Relief algorithm is applied to the weight calculation of feature variables and the optimization of feature factors is realized according to the result of weight calculation.Then,the effectiveness of the IMF energy moment of the multi-channel vibration signals were verified by the K-Means clustering algorithm.Based on the feature factor extraction,the support vector machine model was established for training learning and state prediction,and the SVM-based fault identification system was developed by Matlab GUI.In view of the problem of incomplete faults in the fault diagnosis of the high voltage isolating switch,a diagnostic method is given based on Multi-SVDD to identify unknown,abnormal and known fault types under the incomplete fault category,which enhanced the adaptability of the diagnostic model.The corresponding results showe that the multi-channel vibration feature mergence is more suitable for the isolating switch fault diagnosis.After the feature variables were optimized by the ReliefF algorithm,the accuracy of fault diagnosis can be further improved.