Izvorni znanstveni članak

NEW TOOLS FOR ENHANCED DIAGNOSTICS OF DGA DATA

David Bidwell ; Qualitrol Corporation
Donal Skelly ; Qualitrol Corporation

Sažetak

In the last decade there has been a significant change in the way transformers are viewed. Their
importance together with their obvious value to the network has been enhanced and recognized,
especially in light of the ageing fleet worldwide. At the other end of the spectrum, new transformers are
now being designed and built to tighter tolerances as a result of competitive market conditions, with the
knock-on effect that these “modern” devices do not appear to provide the same stability and longevity as
those that were entering service in the 1970s and 1980s.
Against this backdrop, the advent of transformer monitoring has emerged and continues to
develop at a rapid pace. Although still considered an emerging component of asset management
practice, online DGA is rapidly gaining acceptance and recognition as one of the most powerful tools in
protection against asset failures. While other transformer monitoring technologies abound, many of them
now online, such as partial discharge, these products collectively combine to enable the move to
condition based monitoring of transformer assets.
As online DGA monitors have evolved new products and technologies are reaching the market at
an ever increasing rate. However, the quiet revolution is in the analysis of the data. As more and more
monitors are installed, so the burden of data analysis becomes increasingly large. New ways of extracting
value from this data required. One important approach is the use of Artificial Neural Networks (ANN) for
DGA data analysis. Additionally, with the recognition that data from monitors must be easily transferred
into meaningful information for the end-user, diagnostic tools, such as the Duval Triangle, have evolved
where the addition of Triangles 4 and 5 brings significantly more value to previously mined data.
The mute question in this paper relates to whether or not existing online monitoring hardware has
sufficient accuracy and repeatability of measurement to be of use with these more advanced diagnostic tools.

Ključne riječi

DGA; Online Monitoring; TOAN; Duval Triangles; Artificial Neural Networks

Hrčak ID:

199224

URI

https://hrcak.srce.hr/199224

Datum izdavanja:

15.8.2017.

Posjeta: 1.000 *