An Artificial Neural Network for Pollution Evaluation Based on Leakage Current

An artificial neural network for pollution evaluation is introduced in this paper which contributes to design an expert system for flashover predicting. The network measures four parameters (peak impulse current, maximum low-frequency current, humidity and voltage) which reflect the insulating property of an insulator and outputs the pollution level. After enough of training, the outputs of the network show great agreement with the experiment results. However, accuracy of the network is influenced by the training sample.