Based on Mathematical Morphology (MM)

this paper presents a new method for discrimination between internal fault and magnetising inrush conditions for power transformer protection. The proposed algorithm is to decompose the signal with a class of morphologically simple function

known as Blum functions

to allow the unique feature associated with the inrush current waveform to be preserved

while suppressing other irrelevant features. The proposed scheme has been applied to simulated signals generated by PSCAD/EMTDC. The results confirm that features extracted by the MM filtering have a distinctive property

on which simple identification criteria can be applied for detecting an inrush current from internal fault conditions.