Separation and Measurement Technique Based on Dynamic Frequency-Selection and Frequency Division Applied to Multi Discharge Sources in GIS

Ultra-high frequency (UHF) monitoring technique for partial discharge (PD) is an important technical means to evaluate the insulation deterioration state of GIS equipment. However, huge interference and aliasing of a multiple PD signals in field test effect the sensitivity and reliability of UHF detection. To identify the interference and aliasing of a multiple PD signals more accurately, meanwhile to cover the shortage of amplitude ratio clustering technique which can not identify signals' characteristics in frequency domain, the accuracy and adaptability of clustering algorithm should be further improved. To solve these problems, the separation and measurement technique based on dynamic frequency-selection and frequency division applied to multi discharge sources are proposed. UHF conditioner with high sensitivity and wide dynamic range are designed, whose work modes include wide-band amplification detection and narrowband frequency selection. Accordingly, the interactive dynamic clustering algorithm is proposed. The PD tests in real GIS with five kinds of PD defect and interference models are designed, the real PD tests indicate that the interactive dynamic clustering algorithm is more adaptable in the complex environment of the multiple PD and interference signals.