An Improved Fault Detection Scheme for Power Transformer Protection

In this article, a new algorithm based on time-frequency analysis of differential current is presented for power transformer protection. Since the differential currents are non-stationary signals, the hyperbolic S-transform can be used as a powerful signal processing tool that yields the complete information in both time and frequency domains. A criterion function is proposed based on some extracted features from the obtained hyperbolic S-matrix and frequency contours. By selecting the proper threshold value, the internal fault can be detected correctly from other conditions. Various conditions of internal and external faults, transformer energization, over-excitation, and different levels of current transformer saturation are simulated using PSCAD/EMTDC software (Manitoba HVDC Research Center Inc., Manitoba, Canada), while the important parameters that have a direct effect on the differential current waveform are considered. Current transformers are simulated using accurate Jiels–Atherton model, and a real 230/63-kV power transformer is modeled based on a unified magnetic equivalent circuit. The obtained results show that the proposed algorithm remains stable during external faults and sends a trip signal in less than one cycle in the case of internal fault condition, even when the current transformers are saturated. Also, the effectiveness of the proposed algorithm is verified by using real data obtained from an event recorder of a three-phase power transformer.