Study On GIS Fault Location And Diffusion Of Sulfur Hexafluoride Decomposition Components Based On Computational Fluid Dynamics

The Gas Insulated Switchgear(GIS)is the high-voltage electrical switchgear widely used in places like grid transmission,substations,power plants,and electric power systems.The SF6 insulating gas filled in GIS will decompose under the action of partial discharge or high temperature and generate various decomposition components.By detecting these components,it can be analyzed whether partial discharge or overheating caused by insulation defects occurs in GIS.At present,most detection technologies for SF6 decomposition components are static,that is,collecting sample gas on site and sending it to the laboratory for testing.This method can know whether there is insulation defect inside GIS equipment,but it cannot be clear when and where it appears.With the continuous development of gas sensor technology,the detection technology of SF6 decomposition components will also move towards the direction of on-line and real-time.Firstly,by referring to the literature.this paper introduces three main discharge types in GIS,and gives the possible decomposition products of SF6 under discharge conditions,including SO2,SOF2,SO2F2 and CO2,and puts forward the method based on computational fluid dynamics(CFD).FLUENT software was used to simulate the gas diffusion effect in GIS equipment.Secondly,this paper will use a dynamic perspective to analyze the whole process of SF6 decomposition components from generation to uniform diffusion.On the theoretical basis of Fick’s law and FSG equation,the diffusion process of four main decomposition components of SF6 in high-pressure SF6 gas environment will be studied.The influences of components type,initial gas amount,gravity and fault location on gas diffusion effect were analyzed,and the time required for gas with different diffusion coefficients to diffuse to uniform and the concentration change rate at the monitoring point were explored.Then,on the base of the conclusion,a GIS model with complex internal structure is established,and a large number of simulation operations are carried out to provide data support for fault location.Finally,on the base of the obtained simulation data and according to the statistical learning theory,this paper uses the support vector machine(SVM)to realize the multiclassification of the fault region.By selecting the appropriate kernel function and reasonably optimizing the parameters.according to the component concentration characteristics,the fault location based on the gas diffusion effect is realized,and the positioning accuracy can reach 92.5%.