Frequency response analyses via rational function fitting

Based on terminal measurements on a single phase, low voltage transformer model, it is demonstrated how faults introduced at different positions along the winding could be identified with reasonable accuracy using rational function fitting. Fault in the present context represents a discrete change, e.g. short-circuiting a few turns within a disc (i.e. pre dominantly an inductive change) and the transfer impedances are measured. The proposed method aims at fitting the measured data to iteratively the zeros/poles of the transfer function. Location of such synthesized zeros/poles reveals the nature of fault. Results presented demonstrate the feasibility of this method.