A new statistical approach for analysis of tree inception voltage of silicone rubber and epoxy resin under AC ramp voltage

Abstract

Weibull distribution has been used widely by many researches around the world especially in the analysis of high voltage experimental data. Unfortunately, the statistical techniques used to analysis the high voltage experimental data are not highly accurate. In view of the foregoing, this paper presents a new statistical approach to analyze the tree inception voltage of silicone rubber and epoxy resin. The tree inception voltage of silicone rubber and epoxy resin was measured via camera-equipped online monitoring system. The leaf-like specimen was used as test sample. AC ramp voltage was applied to obtain the tree inception voltage of silicone rubber and epoxy resin. It was observed that, the electroluminescence emission and ultraviolet (UV) radiation occurred indicating the early stage of tree occurrence. The obtained results were analysed statistically by using fitting method. Anderson-Darling goodness-of-fit test was performed in order to obtain the best fitting distribution. Comparison was made between the best-fitted distribution and Weibull distribution. Based on Anderson-Darling tests, the tree inception voltage of silicone rubber and epoxy resin was best fitted with Johnson S B distribution. Based on this fitted distribution, the value of tree inception voltage for silicone rubber and epoxy resin was calculated and equalled to 11.80 kV and 20.11 kV respectively. From this study, it was found out that the best-fitted distribution for the value of tree inception voltage for silicone rubber and epoxy resin is the Johnson S B distribution by means of Anderson-Darling goodness-of-fit test.