Clustering Research on Fractal Parameter of GIS PD UHF Signals

The distribution of fractal parameters (box dimension and lacunarity) vary with noise level of GIS PD UHF signal is the main obstacle to the use of fractal theory in GIS PD pattern recognition. According to the characteristic of fractal parameters, box dimension and wavelet method have been used in control the noise level and gather the fractal parameters of each kind of GIS PD UHF signals. Simulations show that, the fuzzy control parameters filtering algorithm based on the box dimension will lead to the dispersive of lacunarity and the distribution of fractal parameters changes as the noise level of GIS PD UHF signal changes. However, the wavelet method has a good performance in gathering those two fractal parameters of different GIS PD UHF signals. And its a promising approach to expand the applicability of classifiers used in GIS PD pattern recognition.