Monitoring dielectric insulation systems is important for the maintenance of electric assets.In open-air substations, this task can be done with antennas because they cover wide areas and detect the electromagnetic emissions of one of the most common deterioration processes: partial discharges (PD). This detection can be affected by radio-frequency interferences that contaminate the measurements and may lead to errors in the diagnosis. Thus, it is necessary to apply denoising techniques to recover the pulsed signal of this PD activity. The aim of this paper is to propose an adaptive and automatic algorithm named Ensemble Empirical Mode Decomposition(EEMD) which consists in the decomposition of the signal in several Intrinsic Mode Functions (IMF) that will be used to reject UHF interferences superimposed to PD pulses.