Sulfur Hexafluoride decomposition
This paper presents several methods for diagnosing partial discharge (PD) in a Gas insulated switchgear (GIS) chamber using SF6 gas decomposed component analysis. Nine types of purposefully created defects within the chamber were simulated, resulting in PD and hence decomposed byproducts. In contrast to previous studies that only looked at specific aspects of PD analysis, this study employs decomposition component analysis to classify the defect type and assess the severity of SF6 gas decomposition. A series of experiments involving conducting and nonconducting defects were carried out. SF6 decomposition produced SO2F2, SOF2, CO2, SO2, C3F8, SiF4, S2F10, SF4 and CO gases. The byproduct gas concentrations were used as feature parameters in the random forest algorithm to classify single and simultaneous multiple defects types. Next, two SF6 decomposition byproduct concentrations, C(SOF2 + SO2) and C(SO2F2), were chosen as feature parameters for the severity assessment of SF6 decomposition. Three severity levels were then defined using the K-means algorithm. Results show 92.6% defect type classification accuracy is given by the random forest algorithm. Based on the findings, the proposed approaches are advantageous for PD analysis in GIS to ensure its safe and reliable operation.
This results is obtained by the decomposition of sulfur hexafluoride gas under the activity of partial discharge induced by nine insulation defects