Partial Discharge Pattern Recognition For Alternative Liquid Insulation Using Artificial Neural Network

Latif,, Melda and Restina, Anggia (2008) Partial Discharge Pattern Recognition For Alternative Liquid Insulation Using Artificial Neural Network. In: Asian Conference on Electrical Discharge, November 23-25, 2008, Bandung, Indonesia.

Preview

Text P-21.pdf Download (1MB) | Preview

Abstract

This paper report PD pattern recognition in altrnative liquid insulatoin of transformer such as palm oil, soybean oil, and corn oil using artificial neural network (ANN). PD pattern is classified in three kind of pattern, those are No PD pattern, Medium PD pattern, and High PD pattern. Classified PD pattern is modeled using -q pattern. The patterns are learned by the back-propagation method. Three layers are used to optimizing the learning process where hidden layer is varied to get the optimal. In learning process PD data of soybean oil and corn oil are put into the ANN. If the learning process is succeed with indicator that used is the good recognition of ANN, then unknown PD pattern data from some of soybean oil and corn oil are put into the ANN to get the performance. If the performance is good enough recognition, then the network is tested using unknown PD pattern from palm oil. The result shows that ANN can recognise 100% of the training data correctly and over 93.33% of the test data.

Actions (login required)

View Item