Chromatic classification of RF signals for partial discharges and noise characterization

Electromagnetic radiation emissions from partial discharges (PD) can be measured, which could be helpful for avoiding unwanted asset disconnections in the power grid for insulation evaluation. These radio-frequency (RF) measurements do not require galvanic contact with high-voltage (HV) devices, which seems to be a clear advantage, but the lack of synchronization voltage makes PD source recognition a difficult task. Moreover, the presence of RF noise sources in the wideband required for PD detection, can lead to false warnings in the HV apparatus evaluation. In this paper, some specific parameters for signal characterization have been used in three-dimensional maps: signal energy, bandwidth and characteristic frequency. A digital acquisition system has been programmed to measure thousands of emissions from environmental noise and controlled PD sources in laboratory experiments. The aim of this paper is to state if this signal classification technique is reliable for PD and noise characterization in the RF range.