ABSTRACT
Partial discharges (PD) monitoring in HVDC cable systems used for transmission and distribution networks involves very different strategic approaches compared to PD monitoring in HVAC cable systems. PD activity in HVAC cable systems appears when its main insulation has an internal defect (several hundreds of pulses per second can be sustained for hours or days), on the contrary in HVDC cable systems very few PD pulses related to internal defects occur (for example, some units per minute). However, simultaneously there may be many thousands of pulses per minute associated to external phenomena such as corona. PD instruments must be able to discriminate different pulse-type signal sources to reject those that are not related to an internal defect, such as background noise, corona effect, etc. After a different PD cluster is identified next step is to recognize it is due to a defect. As the classic phase resolved PD patterns are not available in HVDC measurements, more complex artificial intelligence tools are required recognize a PD pattern. Numerical sequential series of PD pulses, which amplitudes and time intervals between consecutive pulses, are the only input data to identify what PD defect is involved in this PD source. Powerful noise rejection tools are also required; otherwise, incorrect numerical sequential series will be recorded for the analysis by the artificial intelligence tool and the defect identification may be incorrect.
KEYWORDS
Partial discharges in HVDC, HFCT sensor, PD instrument qualification, transfer impedance, Electrical noise, PD sensitivity, PD clustering, Insulation defects, PD pulse trains, PD patterns in HVDC.
AUTHORS AND AFFILIATION
(3) AMPACIMON (AMPACIMON) (Spain).