Defects in transformer insulation cause partial discharges (PD) which over time can progressively deteriorate the insulating material and possibly lead to electrical breakdown. Therefore, the early detection of the PD is crucial. A PD emits an electromagnetic signal in the ultra high frequency range, and current digital hardware has made it possible to transform this raw data into Phase-Resolved Partial Discharge (PRPD) pattern. Automated PD diagnositc systems previously employed pattern recognition techniques. However, specialists are now able to identify features of the PRPD pattern and deduce different behaviours and therefore physical geometical aspects of the defect. Using this knowledge within a knowledge-based system provides and explanation, and therefore reassurance of the diagnosed fault. This paper describes how to capture and model the knowledge from experts, along with the construction of a rule-based system. It presents a case study of the system's use and the introduction of explanation when diagnosing a defect. The next stage of the monitoring process involves linking the online PD data capture to further diagnostic algorithms and user interfaces. This paper illustrates how this knowledge-based system integrates into an overall transformer monitoring system to provide further data handling and interpretation.
|Number of pages||6|
|Publication status||Published - 2007|
|Event||15th International Symposium on High Voltage Engineering - Ljubljana, Slovenia|
Duration: 27 Aug 2007 → 31 Aug 2007
|Conference||15th International Symposium on High Voltage Engineering|
|Period||27/08/07 → 31/08/07|
- partial discharge data
- knowledge based approach
Rudd, S. E., Strachan, S., Judd, M. D., & McArthur, S. D. J. (2007). An incremental knowledge based approach to the analysis of partial discharge data. Paper presented at 15th International Symposium on High Voltage Engineering, Ljubljana, Slovenia.