One of the standard tests carried out on a power transformer is the lightning impulse, to corroborate the integrity of its winding insulation. The existing transfer function method used for diagnostics lacks sensitivity in detecting minor internal faults. Hence a new approach using 'wavelet transforms' is developed to analyze the non-stationary neutral current signal which varies due to the existence of a fault. This approach involves the choice of a basis function applied on the neutral current signal and analyzed in the time-frequency domain. The effectiveness of the approach is that the exact instant of fault occurrence is directly known, such that its actual location in the winding can be predicted easily. This paper presents the approach to simulated interturn fault at three selected locations in the partially interleaved HV winding of a 50 MVA, 400/11.5 kV station transformer such that its performance is recognized. Minor faults, such as one turn short, is simulated and successfully detected using this approach. The results are in good conformity with the theoretical calculations based on the length of the conductor involved and the wave propagation velocity.
展开▼
