Realtime fault detection and classification in power systems using microprocessors

Abstract

A method is proposed for real-time fault detection and classification in power systems. The method can be implemented in low-cost units based on 8-bit microprocessors. These units are part of a multiprocessor system which allows the detection of single line to ground faults, double line and double line to ground faults, and three-phase faults. The detection of an overcurrent triggers two simultaneous processes executed in real time: the waveforms of the voltage and current signals are registered and a probabilistic criterion is used for fault confirmation. The process of confirmation is based on the estimation of the three phase voltage phasors by means of a set of Kalman filters, and on the calculation of the fault probability. If the fault has been classified correctly, a recording is made with the samples acquired, including five cycles of the prefault signals and 150 cycles of the postfault signals. This recording is use by the multiprocessor system to calculate the fault distance and to generate an accurate fault recording. The algorithm has been verified using a computer simulation of the different types of faults, and has been optimised so as to detect and correctly classify faults in less than 20 ms.