Research On Condition Monitoring And Fault Diagnosis Method Of Gas Insulted Switchgear

With the development of the power grid of the electrical system in our country,the application of high voltage equipment is more and more extensive.Because of its compact structure,strong anti-interference ability,high reliability Gas Insulated Switchgear have been widely applied in high voltage substation and ultra-high voltage substation.However,in recent years as the growth of the equipment running time and rugged operation condition in substation,the GIS equipment failure occurs frequently.In this way,that state monitor and fault diagnosis method of GIS switch equipment is studied,which can help the field staff to discover the fault early and arrange the overhaul in time so as to ensure the reliability of system operation.On the basis of consulting relevant literature,this paper makes a detailed study and analysis on the basic structure and common fault mechanism of GIS,and selects monitoring variables that can reflect the operating state of equipment based on common fault mechanism.Aiming at the frequent problems of GIS equipment fault in substation site,a GIS equipment fault diagnosis algorithm framework based on state estimation and residuals classification is proposed.The algorithm includes three parts: establishment of state estimator,fault warning and fault recognition.the concrete implementation steps are as follows:(1)In view of that complex structure of GIS,the traditional multi-input single output support vector regression cannot be accurately modeled,and a multi-input multioutput support vector regression is established by improving the model structure to establish the equipment state space model,at the same time,the optimization method of model parameters is given.(2)Using the status estimator to estimate the state variable of the device,and comparing the real-time measure value with the estimated value to calculate the residual error of the multi-state variables and the index of the fault.In order to avoid model errors and noise interference,adaptive threshold method is selected to detect GIS equipment faults.Once detected faults,alarm signals will be given.(3)Soft-max regression classification model is trained based on the actual measurement data in the field.After the fault warning,the trained classification model is used to classify the residuals of multi-characteristic variables to identify the fault types,and finally realizes the GIS equipment fault diagnosis.Under the environment of MATLAB,the particle filter,improved support vector regression,adaptive threshold and regression classification algorithm are programmed.A simulation analysis is carried out based on the measured data of substation,and the simulation results have verified the validity of the diagnostic method of GIS that is consist of the particle filter and soft-max logical regression.