Denoising of the neutral current using wavelet for diagnosis of the transformer

Abstract

Insulation failure within transformer winding is considered to be one of the most important causes of failure of the power transformer. Comparison of neutral currents at reduced and full voltages is an important criterion for detection of fault in the power transformer winding during impulse test. Any difference in wave shape of the recorded current and voltage signal shows the failure of transformer. In this paper, identification and separation of a noise signal from the main neutral current are carried out using wavelet technique. In order to carry out analysis faults are created in discs of the windings of a 61 MVA, 11.5/230 kV generator transformer at different locations. Neutral currents are recorded by applying low impulse voltage. Contribution of each frequency in time domain signal is calculated by Daubechies wavelet. The proposed method can be used for isolating the contribution due to any noise frequency in the waveform.