Application of probability distribution to the measurements of partial discharges deriving from tree-growth tests

Statistical tools are applied to the results of partial discharge measurements performed by means of a digital system, with the purpose to identify parameters useful for pattern recognition. In particular, the probability distribution of the charge height is investigated resorting to the two-parameter Weibull function. The PD data obtained from measurements of tree growth in specimens of EVA copolymer subjected to constant voltage, at room temperature, are considered. Two configurations of the point-plane electrode geometry are tested, which differ for the presence or not of an artificial void at the needle tip. It is shown that the probability distribution of the PD-pulse height is well described by the two parameter Weibull function when tree grows in the insulation, in the absence of the artificial void, or PDs are occurring in the void, before inception of tree in the insulation. The Weibull parameters are related to tree-growth features and are sensitive to transition of tree from void to insulation