A Multi-source Partial Discharge Signals Separation Method Based On UHF Envelope Detection Waveform

In recent years, the UHF partial discharge (PD) detection has been widely used and become one of the primary means of high-voltage electrical equipment condition monitoring. However a lot of practices confirm that the complex electromagnetic interference environment, especially the exits of multi-PD sources, remains a serious bottle neck for partial discharge accurate detection and diagnosis.In order to effectively solve complex interference problem in electrical equipment PD detection using UHF, especially for the pulse discharge interference, this paper presents a method for multi-PD source signals separation, based on UHF envelope detection waveform. The PD UHF signal data is collected using a high-speed oscilloscope. The UHF envelope detetion parameter is studied though software emulation and a UHF envelope detector conditioner is developed on this basis. The typical PD model tests have been conducted to study the influences of different types of defects and different propagation paths to the waveform characteristics of PD UHF envelope detection. The envelop detection waveform characteristics are constructed and screened, the waveform similarity coefficient and integral area/peak are eventually indentified as waveform clustering parameters. Multi-PD source separation experiment has been performed on the GIS test platform and researches on the clustering effect of PD sources based on the2-dimensional scatterplot of waveform similarity coefficient and the integrated area/peak has been done. The results indicate that UHF envelope detection waveform can make effective separation of Multi-PD sources. Effective clustering algorithm of UHF edvelope detection waveform is studied and A UHF PD detecion and automatic clustering software is developed based on labview platform.Systematic studies indicate that UHF envelope detection waveform can make effective separation of Multi-PD sources, which provides a new solution for accurate diagnosis using UHF with Multi-PD sources exist.