Diagnosis and Protection of Stator Faults in Synchronous Generators Using Wavelet Transform

This paper presents a novel application of wavelet transform for the diagnosis and protection of stator faults in synchronous generators. The instantaneous powers of the wavelet packet transformed coefficients (dd²) of line voltages and currents decomposed up to the second level of resolution of the wavelet packet tree using a mother wavelet are used to detect and differentiate different faulted conditions from normal (unfaulted) conditions. The performance of this newly devised protection technique is evaluated by simulation results as well as by experimental results. The complete protection scheme incorporating the wavelet power based detection algorithm successfully implemented in real-time using the ds1102 digital signal processor board for a laboratory 1.6 kW three-phase synchronous generator. The stator phase unbalance, single line to ground (L-G) fault, line-to-line (L-L) fault and turn-to-turn fault are investigated in order to evaluate the performance of the protection technique. In order to prove the superiority of the proposed protection technique over the conventional techniques, a comparison between the proposed and the conventional discrete Fourier transform (DFT) based diagnosis scheme is made at different dynamic operating conditions.