Transient stability and critical clearing time classification using neural networks

The paper presents a novel AI based artificial neural network (ANN) classifier for AC interconnected power system on-line dynamic first swing stability assessment and classification of critical clearing time as either short (below 50 ms) or long from (50 ms-200 ms). The classification is done using a multi-layer logsigmoid activation and back error propagation (BEP) for training and weight adjustment. The detection scheme is based on a discriminant vector of FFT spectra of synchronous generator rotor angle, speed deviations, and accelerating power magnitude and their cross correlations power spectra