AN ADAPTIVE SCHEME FOR DIGITAL PROTECTION OF POWER TRANSFORMERS

This paper presents the development of an adaptive scheme for digital differential protection of power transformers. First, the transformer status is monitored to decide if the transformer is disconnected or is in normal operating condition. When the transformer is energized, a two-state Kalman filter model monitors the transformer during normal operating conditions. When a transient is detected, a three-state Kalman filter model is initiated to obtain the best estimate of the phasor quantities in the shortest period of time. These phasors are used to decide if a fault in the transformer exists. If the transformer was disconnected and switched on, the current waveform would include inrush current or a combination of inrush and fault currents. In this case a five-state Kalman filter model is initiated to obtain the best estimates of the dc offset, fundamental and second harmonic phasors. These phasors are used to decide if the switching occurs in the presence of an internal fault. Second, the current transformer ratios and tap values are included in the algorithm to obtain an adaptive percentage differential characteristic. This adaptive scheme was tested on data obtained in the laboratory and proved to be highly accurate and fast.