Insulation condition assessment of 11 kV paper cables using neural networks

The insulation condition of 11 kV paper cables was analyzed using self-organizing maps, a type of artificial neural networks. The cable was energized at voltages above its nominal level in order to degrade the cable and obtain the discharge activity at different condition states. This process was repeated several times until a final breakdown on the sample was obtained. The result obtained for all the trained maps confirms that the data obtained before the breakdown came from a fatal condition of the insulation