Discrimination of partial discharge sources in the UHF domain

Partial discharges (PD) are a significant contributor to insulation failure within power transformers, and as such, the localizing of a PD source can be of critical importance when it comes to a transformer's assessment and repair. One PD localization technique which is acquiring increased attention is the application of an ultra high frequency (UHF) sensor to trigger an acoustic array. This hybrid approach helps to overcome the poor signal to noise sensitivity issues commonly associated with externally mounted acoustic sensors by leveraging the benefits offered by an internally mounted UHF sensor to trigger PD events. The internally mounted UHF sensor is insensitive to external corona discharges which typically are the largest disturbance source associated with conventional measurements. A problem with this hybrid methodology is when there are multiple PD. The research presented in this paper demonstrates that UHF PD signals are primarily independent of the type of PD source. The research also demonstrates that UHF PD signals inherit signal characteristics which are particular to their location within a transformer structure, and inherit distortion characteristics which are particular to their journey from source to destination. As a result it is proposed that the correlation of UHF PD signatures can be used for discrimination purposes in order to identify individual PD sources within a transformer using a single UHF sensor. This could then be used in conjunction with acoustic measurement techniques to sequentially localize multiple PD within a power transformer which formerly has proven to be a technical challenge.