Neural networks application to the fault location in the transformer winding

Transformers are usually subjected to lightning impulse tests for their withstand ability after assembly in the manufacturing plants. Unexpectedly, the windings are sometimes broken down by excessive voltage stress due to internal oscillation. Naturally, the location of the damaged portion, which might take a long time, has to be found for repair work. If location is easily done, subsequent jobs will be effectively and quickly handled. This paper shows the possibility of neural networks application to the fault location in the transformer windings, where short circuit failure is assumed to occur between some coil sections.