Research On Online Monitoring Of Mechanical Characteristics And Insulation Status Diagnosis Of 10kV Ring Main Unit

Ring main unit is an important electrical switchgear in distribution network.It is responsible for adjusting the connection mode of distribution network lines to ensure the safety of power supply.In the operation of ring main unit,mechanical and insulation faults may occur,which will affect the normal working state of distribution network.In order to better monitor the working status of the ring cabinet,it is necessary to study advanced technical methods to monitor the mechanical characteristics and insulation status of the ring main unit.Aiming at the mechanical characteristics and insulation status of ring main unit,the research object of this paper is XGN-12 air-insulated ring main unit produced by Beijing Shuangjie Electric Company.The mechanical characteristics test platform and the discharging sound signal test platform are built respectively.A data acquisition system based on LabVIEW is designed to collect,save and process the mechanical characteristic parameters of the ring main unit and the discharging sound signal of the internal insulator device.Acceleration signal,displacement signal,permanent magnet coil current and fracture signal are collected through mechanical characteristic test platform.In the insulation state test,different levels of voltage values are generated using the test transformer.In the dry or humid environment,the insulator in ring main unit is discharged,and the discharge visible sound signal is collected.In the analysis of mechanical characteristics,aiming at the acceleration signal in the process of switching-on and closing,the interference is eliminated by low-pass filter pretreatment,and then the acceleration signal is decomposed by wavelet modulus maximum method to obtain 5 or 6 layers of high frequency components.The breaking point of the signal is found in the high frequency components as the completion time of switch opening and closing,which is the break jumping time.The opening and closing time is calculated,and the mechanical characteristic parameters such as travel,over-travel,average opening speed and average opening speed are calculated with displacement sensor.Wavelet analysis is used to extract the time of break in closing and closing,which solves the problem that it is difficult to obtain the signal of break in on-line operation of ring main unit.The insulation status of ring main unit is detected by analyzing the discharge sound signals of insulators in ring main unit.The short-time energy method and adaptive notch filter are used to preprocess the signal.Then the feature vectors of the energy distribution of the discharge sound signal from 6kHz to 18 kHz are obtained by wavelet packet analysis.The 6 dimensional energy distribution feature vectors are obtained by Fisher’s linear dimensionality reduction criterion.Finally,the energy distribution feature of discharged sound are trained by using two algorithms,Gauss naive bayesian(GNB)and one class support vector machine(OC-SVM).The recognition effect of OC-SVM is better than that of gauss naive bayesian classifier by comparing the recognition effect of the two machine learning algorithms.By analyzing the discharge sound signal,it can satisfy the qualitative analysis of the insulation state of ring main unit in low cost and complex environment.