Prediction of transient stability status using Walsh-Hadamard transform and support vector machine

Abstract

A novel scheme is presented for real-time prediction of rotor angle stability status. The scheme is activated following a large disturbance and operates by obtaining a set of two rotor angles for each generator. Each set of two rotor angles sampled is decomposed using the fast Walsh-Hadamard transform resulting in two Walsh coefficients. The absolute value of each coefficient is obtained and the maximum of the two is selected. The selected absolute maximum values, one for each generator, are then summed and fed into a trained support vector machine which predicts the stability status. The scheme was tested using the IEEE 39-bus system. The prediction accuracy of the scheme proved to be 100%.