Searching for indexes suitable for rotating machines diagnosis

Results of investigations having the purpose to select meaningful diagnostic indexes for the identification of different defects of insulation systems of AC rotating machines are reported in this paper. Different discharge sources were created artificially on six Roebel bars arranged in a mock-up of eight stator slots. These bars have been firstly tested one by one, then connected in series to simulate different locations of PD sources. Standard quantities and parameters, derived from PD-pulse height and phase distribution analysis, were compared in order to verify their ability to identify the defects in the various set-up conditions (bars alone or connected together, changing couplers and instrument bandwidth). It was found that parameters related to the shape of PD patterns remain almost constant and their suitable processing allows identification of the different PD sources