Discrimination of partial discharge patterns using neural network

Describes the application of a neural network algorithm to the perception of partial discharge patterns. The aim of this preliminary research is to see if the neural network is able to discriminate the tree initiation from the needle-shaped void gap. A needle-shaped void sample was made from epoxy resin to generate an electrical tree under AC voltage. The partial discharge patterns before and after the tree initiation were learned by the neural network using the back-propagation method. After the learning process was over, the unknown discharge patterns were put into the network. The network demonstrated good performance with regard to discriminating the tree initiation. It was required for stable discrimination of tree initiation that the tree length be larger than the extent of the void size