ANN-based Improved Zone-3 Decision

Early detection and discrimination of symmetrical fault from non-fault transients is the need of hour for reliable operation of power system. Incorrect Zone-3 decision may cause mal-operation of relay and subsequent cascading outages. To mitigate this, an artificial neural network (ANN) based Zone-3 supervision scheme is proposed in this paper. The proposed algorithm comprises of two stages. In Stage-1, ANN architecture is proposed to estimate the current samples. The reconstructed samples are compared with actual samples to detect for any transients in the line. In case of detection of transients, Stage-2 is employed to compare the phase angle of the positive sequence impedance with a threshold. The threshold is selected such a way to segregate fault from stressed conditions. Various stressed and fault conditions are simulated using PSCAD/ EMTDC software. The proposed method is tested in MATLAB environment and significant results are achieved.