Investigation of partial discharge in single void and multi-voids using data mining technique

Partial discharge (PD) is a phenomenon that commonly occurs in the insulation of power system equipment. In this paper, PDs occurring in a single void and multi-voids were investigated. Air voids at atmospheric pressure were injected into cylindrical epoxy resin blocks to create test samples with single void and multi-voids. PDs from the samples were recorded and further analyzed using phase resolved analysis. A data mining technique using decision tree (DT) and genetic algorithm (GA) approach was then applied to mine the rules/relationships which can be used to differentiate PD from single and multi-voids. Also investigated were the PD characteristics as the number of voids existed in the solid insulation increased.