The paper considers the diagnosis of HV constructions which suffer fro
m partial discharge (PD). Using previously developed statistical techn
iques the properties of discharge patterns are quantified in a fingerp
rint. Through the analysis of different defects and their fingerprints
in HV constructions, the practical implication of these techniques is
discussed. Furthermore, different possibilities of developing a data
bank for the recognition and diagnosis in HV components are shown.