Study Of The Diagnosis Method Based On Image Of Creeping Discharge

The interface between the solid dielectric insulation and gas is a weak link of high voltage insulation systems. Creeping flashover occurred frequently, which caused a immeasurable loss to the power sector. How to accurately diagnose the creeing discharge becomes the key to reduce the loss. This requires a comprehensive understanding of the development process of creeping discharge and the factors affecting the development of creeping discharge. Therefore, study of creeping discharge has become a major issue. This paper studied surface discharge under different physical conditions by using digital image processing technology.A new method based on the studying of colorimetric properties of creeping discharge image was proposed to extract the effective discharge area. By analyzing the characteristics of each color component(R?G?B)of creeping discharge:they have different ability about reflecting the discharge area,component G reflects the main part of the discharge,with relatively independent distribution area.Component R and B can reflect the edge area of the discharge,and they have similar distribution area.Then we combined the gray color conversion formula with RB,s similarity to deduce the range of R+B and G by combining R and B as a whole(R+B),thus we can determine the background threshold to extract the effective discharge area.By studying the grayscale image statistics index, fractal index and UHF signal characteristics of creeping discharge indicated that the slide flash phase consists of two phases: the pre-slide flash phase and the slide flash phase. This achievement will help to deepen the understanding of the phenomenon of slide flash, specific content is as follows:(1) grayscale image statistics index: when creeping discharge turns to pre-slide flash phase, the discharge distance increased sharply, standard deviation and coefficient of variation's head features disappears while increasing dramatically so that forming a characteristic peak in middle region. After further development to the slip flash phase, discharge distance increase again, standard deviation and coefficient of variation's amplitude decrease while the central peak of coefficient of variation develops to the rear.(2) fractal index: the fractal dimension maintains a low level and has low space occupancy at the initial phase. After pre-slide flash phase, fractal dimension increases significantly, and maintains a moderate increase with increased voltage. After the slide flash phase, the fractal dimension increased sharply again. Its ability to fill the space near the two-dimensional.(3) ultrahigh frequency index: UHF signal consists of time domain and frequency domain signal. Time-domain signal: its UHF pulse amplitude increases significantly at the pre-slide flash phase and the slide flash phase. Frequency-domain signal: creeping discharge frequency domain mainly in A(0~150MHz), B(500MHZ~750MHz), C(800MHz ~1.7GHz). The intensity of the band A is strong while band B and C is almost 0 at the initial phase. After the pre-slide flash phase, band B, C significantly enhanced and continue to enhance with increasing voltage. After the slide flash phase, band B, C increase sharply again.