Abstract:
Transformer are cardinal part of power system and costly device it is responsible for the persistence of the supply any discrepancy on transformer can led to the power failure thus it need to be well protected from all types of faults. Mostly differential protection is employed for the protection of the transformer because it provides the overall protection of transformer from all types of internal fault .Discrimination between inrush and internal fault is always a tedious task for the protection of transformer as modern transformer protection failed by conventional harmonic based thresholding techniques therefore new intelligent methods are proposed now a days for the protection of Transformer. Earlier techniques where based on the threshold and these techniques where harmonic restraint methods comparing different harmonic ratios, now these techniques not applicable because due to the modern core transformer with high permeability and low coercion core material. For the detection of internal fault this dissertation proposed the algorithm consist of the two techniques of the artificial neural network i.e., radial basis function and back propagation compare with the support vector machine .We have classified internal fault from the other four conditions of the transformer .The dataset required for simulation is formed by the simulation done on the PSCAD with the real time simulation data available from the Jabalpur electricity board in which we perform the simulation for active and reactive power in all the above conditions of the transformer.
