PD pattern recognition with neural networks using the multilayer perceptron technique

Abstract

The partial discharge (PD) pattern recognition capability of a neural network, employing the multilayer perceptron technique with data input based on five discharge pulse form parameters, is examined. Simple discharge sources, consisting of artificially created cylindrical cavities with metallic and dielectric electrodes, are employed. The PD pattern discrimination capability is tested using cavities of equal depth but with different electrodes, and cavities of varying depths but with similar electrodes. Preliminary test results are positive. >