Thesis (M.Sc.) - Cairo University - Faculty of Engineering - Department of Electrical Power and Machines

Power transformers are essential elements in power systems and thus their protection schemes have critical importance. In this thesis, an online scheme is developed to observe the operating conditions of the connected transformer to assess the internal fault conditions. The proposed scheme depends on the relation (xV- Iin) which has the ellipse locus. According to the type of the fault, the ellipse dimensions vary and the scheme can effectively extract new features in order to classify and locate the deformation inside the transformer. Two proposed approaches are developed and applied to deal with internal insulation failure problems within power transformer windings. The first approach is applied to classify five different internal insulation faults: inter disk, series short circuit and shunt short circuit, by applying artificial neural network with a reasonable accuracy. The second approach aims to find the exact location of the three types of internal faults along the transformer winding by dividing the transformer winding into sections. Finally, the superiority of the proposed scheme to accurately discriminate and locate the power transformer internal faults is extensively examined by comparing its performance with some published schemes. Thus, it is concluded that it can be applied as a useful tool for condition assessment of transformers enabling power management system to spot the ones requiring immediate periodic maintenance or exchange without supply interruption

Issued also as CD