This paper presents a novel approach based on an artificial neural network (ANN) for identifying the insulation defects of power transformers, which is so-called three-dimensional (3-D) partial discharge (PD) patterns recognition. First, four epoxy-resin power transformers with typical
insulation defects are purposely made. These transformers will be used as the experimental models of PD examination. Then, to establish a database of PD patterns, a precious PD detector is used to measure the 3-D (Φ-Q-N) PD signals of these experimental models in a shielded laboratory.
The database is used as the training data to train a three-layer Back-propagation neural network (BPNN). In this work, a feature extraction method is adopted to reduce the number of dimensions of PD pattern. Moreover, a fast learning algorithm is used to speed up the training process. The
training-accomplished BPNN can be a good insulation defects identification system for epoxy-resin power transformers. The proposed approach is successfully applied to practical epoxy-resin power transformers field experiments. Experimental results indicate that the proposed ANN-based approach
is a powerful and accurate tool in terms of power transformers insulation defects identification. Moreover, the proposed approach has a good tolerance of noise interference.
ADVANCED SCIENCE LETTERS is an international peer-reviewed journal with a very wide-ranging coverage, consolidates research activities in all areas of (1) Physical Sciences, (2) Biological Sciences, (3) Mathematical Sciences, (4) Engineering, (5) Computer and Information Sciences, and (6) Geosciences to publish original short communications, full research papers and timely brief (mini) reviews with authors photo and biography encompassing the basic and applied research and current developments in educational aspects of these scientific areas.