Advances in data mining for dissolved gas analysis

Abstract

This paper reports NGC's continued application and refinement of a data mining technique based on the Kohonen neural network. The technique has been applied to NGC's database of transformer dissolved gas-in-oil analysis (DGA) measurements for high voltage transformers.The technique has proven able to highlight bad data and 'blind test' data, and has been optimized to reveal the early stages of potential plant problems. A number of key types of transformer condition have been distinguished by it, including for example three kinds of partial discharge.The Kohonen technique has been successfully applied to transmission, distribution and generator transformers. In addition a practical tool for DGA interpretation is being developed. We are now looking to expand the use of the technique to other monitored parameters.

Transformer failure diagnosis by means of fuzzy rules extracted from Kohonen Self-Organizing Map

“…Among Artificial Neural Networks, the Kohonen Self-Organizing Map (SOM) has some important features which have led to the increase in its use for transformer fault diagnosis [13][14][15][16][17][18]. The SOM is a pattern recognizer and has been applied largely due to its ability to knowledge discovery from a database.…”

Data mining on partial discharge signals of power transformer's defect models

“… proposed artificial intelligence algorithms, such as the back‐propagation neural network, the k‐means clustering, the C5.0 rule induction, and the Sammon map for the analysis of condition monitoring data, and demonstrated this method in its discovery of useful knowledge from trip coil data captured from a population of in‐service distribution circuit breakers and empirical UHF data captured from laboratory experiments simulating PD defects typically found in high‐voltage transformers. Esp and McGrail reported that National Grid Company (NGC) continued application and refinement of a data mining technique on the basis of the Kohonen neural network. The technique has been applied to NGC's database of transformer dissolved gas‐in‐oil analysis measurements for high‐voltage transformers.…”

A novel approach (scoring system) for integrating dissolved gas analysis results into a life management system

“…One of the most effective investigated by National Grid has been the 'Kohonen net' cluster analysis technique [7] which has been relatively successful in grouping together DGA results of an apparently similar significance and identifying when a result appears to move to a different state. Although such techniques can be argued to provide a better basis for identifying an unusual condition than the purely statistical approach described above, they still rely very much on the human expert to ascribe some significance to the clusters identified.…”