Classification of surface condition of polymer coated insulators using wavelet transform and neural networks

Polymer coated insulators are used to suppress the pollution phenomenon on high voltage insulators. In order to investigate surface activity, which is directly linked to insulation performance and lifecycle, leakage current waveforms are used. In this paper a pattern recognition system is proposed capable of identifying four different types of surface activity. Patterns are extracted from leakage current waveforms through wavelet analysis, and especially multi resolution signal decomposition technique. The identification process is automated using an artificial neural network.