ANN BASED PATTERN-CLASSIFICATION OF SYNCHRONOUS GENERATOR STABILITY AND LOSS OF EXCITATION

The paper presents a novel Artificial Intelligence (AI) based Neural N etwork (ANN) pattern classification and on-line detection scheme for a single machine infinite bus system. The proposed on-line relay and dy namic pattern classifier utilizes specific frequency spectra of the hy perplane discriminant vector of machine rotor angle, speed, accelerati ng power, instantaneous power, voltage, and current using either a per ception single layer detection scheme or a two layer feed forward ANN for on-line classification and detection of fault condition causing fi rst swing transient stability or loss of excitation. Other relay binar y outputs include fault type and allowable clearing time identificatio n. The detection accuracy is improved by utilizing the cross spectra o f discriminant vector input variables correlations. The proposed patte rn classification technique can be extended to interconnected multi-ma chine systems by using relative rotor angles, frequency deviations, ti e-line powers, and their cross spectra variables.