A new Kalman filter approach to digital relaying

Abstract

A novel Kalman filter approach to digital relaying is presented, in which the traditional application of simple, low-cost analog prefilters is assumed. The sampling rate is not directly coupled to the filter bandwidth; while this one was specified as low (150 Hz at maximum), the sampling period was determined by time resolution and processing cost requirements (1 kHz). To cope with the signal distortion resulting from the prefilters' own transients, a novel observation model is developed for the Kalman filter, with very little extra processing. The coloring effect on the observation noise by prefiltering is modeled in the Kalman filter state space, assuming white input noise, but the simulation tests with a modal-based impedance estimator show that little improvement is achieved with that. Instead, the addition of a colored time-decaying sinusoid to observation white noise in the state space supports improved performances, especially for very noisy faults in long UHV transmission lines with weak sending buses. Time decisions between 1/3 and 1/2 cycle were achieved for faults at the ending bus and at mid-line, respectively.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>