A novel method for large turbine generator protection based on wavelet transformation

The frequency property of wavelet transformation (WT) has been studied in depth in this paper with comparison to the discrete Fourier transform (DFT). The band overlapping and frequency leakage among different scales of WT is discussed. Several new wavelet functions with compact-supported property in frequency domain are constructed. A general process for selecting proper wavelet function according to practical requirements is proposed. A novel WT based method is developed to detect generator inner faults. Comparing with the DFT method, the WT based signal analysis requires no frequency tracking and is not affected by frequency deviation. No leakage of frequency occurs with the new method applied. Simulation results prove the feasibility and advantages of the new method presented by the paper.