Skip to Main Content
In order to solve the problem of mis-operation of transformer differential relay owing to the inrush current, internal faults and inrush current must be discriminate effectively. In this paper, a novel approach of identification between inrush current and internal faults based on hyperbolic S-transform (HST) which is a very powerful tool for nonstationary signal analysis giving the information of transient currents both in time and in frequency domains is presented. The signal is transformed to phase space by using HST and the features are detected. It is found that the time-frequency contours in case of inrush current are different from that in case of internal faults. The results obtained by using HST and discrete wavelet transform (DWT) were compared. Comparison results indicate that the time-frequency localization characteristics are more distinct in ST domain, and HST has strong capability in noise reduction. The spectral energy and standard deviation from the HST of signal are computed, classification of inrush current and internal faults is done by BP neural networks. Simulation results indicate that this technique is effective and feasible.