Research On Mechanical Fault Diagnosis Technology Of Open-type Disconnector Mechanism

In recent years,the rapid development of State Grid Corporation of China,substation construction and the number of high-voltage disconnector equipment has increased exponentially,and affected by operating conditions,environment and other factors,prone to corrosion,jamming,inappropriate breaking and closing,refusing to move and other problems,in serious cases will cause equipment failure,line tripping.Along with the popularization and application of big data,cloud computing,Internet of Things,mobile Internet,artificial intelligence and other technologies in the electric power system,it is of great value to carry out research on mechanical fault diagnosis technology of high-voltage disconnector to realize comprehensive perception and intelligent diagnosis of operation status for intelligent operation and maintenance of disconnector.In this paper,by building a test platform for high-voltage disconnector fault diagnosis and testing,simulating typical faults such as jamming,inappropriate breaking and closing,and three-phase non-simultaneous,and combining with methods such as correlation coefficient analysis,feature selection and machine learning,intelligent diagnosis of disconnector faults is realized.The specific research contents are as follows:(1)The typical structure of disconnector is briefly introduced,and the relevant state parameters that can reflect the mechanical state of disconnector are sorted out with the current research status,and the diagnostic principles of each state quantity are elaborated separately.Secondly,the state parameters related to disconnector faults are compared and analyzed,and the torque signal and angle signal are selected as typical state parameters,and the most common GW4 outdoor high-voltage disconnector is taken as an example to analyze the relationship between its mechanical state and torque and angle signals.(2)The torque and angle measurement test platform of the GW4-126 type outdoor high-voltage disconnector was built,and the disconnector mechanical state test unit containing power module,measurement module,control module and power module was designed.At the same time,three kinds of faults,such as isolation switch jamming,breaking and closing not in place,and three-phase non-simultaneous,are simulated to obtain their torque and angle change laws,and the correlation between them and the mechanical state of isolation switch is summarized and analyzed.(3)A fault diagnosis method of high-voltage disconnector based on the RSXGBoost is proposed.First,the correlation between high-voltage disconnector faults and feature quantities such as torque and rotation angle is analyzed by the Spearman algorithm,and the feature quantities are screened by the RFECV algorithm.Then,the high-voltage disconnector fault model is established based on random search optimization with the XGBoost algorithm by combining the screened feature sets.Finally,the RS-XGBoost algorithm is trained and tested by constructing a sample dataset of high-voltage disconnector based on test data.The results show that the average accuracy of the RS-XGBoost model is 93.4%,which can effectively realize the intelligent identification and fault diagnosis of the mechanical state of highvoltage disconnector.