Vibration pattern recognition and classification of electric generator in power system using wavelet analysis

This paper presents an effective approach for multi-concurrent fault diagnosis based on integration of fractal exponent wavelet analysis and neural networks. The characteristics of a signal both in the time and frequency domains can be localized accurately by the wavelet transform. Considering the inter relationship of wavelet transform between fractal theory, the whole and local fractal exponents obtained from wavelet transform coefficients as features are presented to extract fault signals, which are inputted into radial basis function for fault pattern recognition. The network structure and parameter identification are fulfilled by establishing the fault diagnosis model of electric-generator set and using the genetic algorithm. The faults are input into the trained wavelet network by choosing enough samples to train the fault diagnosis network and the information representing. Also the type of fault can be determined according to the output result. This paper discusses the robustness of exponent wavelet network for fault diagnosis. This method can make the practical multi-concurrent fault diagnosis for stator temperature fluctuation and rotor vibration accurate and comprehensive.