Fast transformer fault classification by means of recorded protection signals

The operation status of a power transformer is critical for power system monitoring. This paper presents two methods for power transformer fault classification based on its protection signals. The first method is based on a model of relationships between fault and protection action, which is formulated as a matrix function. A pseudoinversion is used to diagnose the system yielding indices similar to the probabilities of occurrence of different faults. Statistical data on faults and transformer protection operation serve as the basis of the second method that uses Bayes´ probability calculation. The results of both methods are compared and they yield similar results. The proposed methods could serve for fast identification of a power transformer fault condition within system monitoring in the control center