Signature analysis for leakage current waveforms of polluted insulators

Previous researchers have shown, using FFT spectrum analyzers, that arcing causes a significant increase in the third harmonic content of the leakage current of polluted insulators. This paper investigates the feasibility of using modern high resolution digital spectral analysis methods to obtain better spectral estimates. Techniques discussed include the classical periodogram, correlogram, autoregressive (AR) and autoregressive-moving average (ARMA) methods. The leakage current was continuously acquired and stored (for about 30 mins) during flashover tests. The analog waveform was then digitized off-line and the digitized samples used to compute the power spectral density (PSD). It was observed that the relative magnitudes of the odd harmonics in the leakage current increase with increase in arcing intensity. In particular the third harmonic PSD increased from almost zero at the start of the experiment to about 20% of the fundamental frequency just prior to flashover