Partial discharge mechanism detection by neuro-fuzzy algorithms

The study of the partial discharge (PD) activities allows to highlight the ageing stage in the insulation material under test. During the life of the insulating material, the shape of PD signal changes relating to the ageing stage. This paper proposes an analysis on a dielectric system, affected by partial discharge activities, based on an original processing of the acquired data. This approach allows to minimise the necessary classification data. In what follows the basic steps of this approach are reported: data acquisition; identification of the significant parameters using wavelet transformed signal; fuzzy processing of the data. The main results presented in this paper are: a full automatic process; a diagnostic index of the phenomenon trend; if is possible to consider this index as a way to monitor dielectric behaviour. A short description of the phenomena, together with an introduction to wavelet and fuzzy logic are reported too.