Diagnosis of Partial Discharge Using Self Organizing Maps and Hierarchical Clustering - An Approach

This paper shows a first approach in a diagnosis selecting the different features to classify measured of partial discharges (PD) activities into underlaying insulation defects or source that generate PD. The results present different patterns using a hibrid method with Self Organizing Maps (SOM) and Hierarchical clustering, this combination constitutes an excellent tool for exploration analysis of massive data like partial discharge on underground power cables. The SOM has been used for nonlinear feature extraction. Therefore, the clustering method has been fast, robust, and visually efficient.