Abstract
The dissolved gas analysis (DGA) technique is widely accepted in identification of incipient faults for power transformer. However, the diagnostic accuracy of classical DGA methods is compromised by absolute code boundary or noise-corrupted data. To resolve these problems, this paper presents a novel scoring method based on strong correlation gas ratios, weights, and effective interval under medium-sized data samples. By means of the ratios selected by coefficient of variation and calculation of weights by technique for order preference by similarity to an ideal solution, a score formula can be built. Taking the ratios as the input of formula, the score can be calculated and the specific fault can be identified by looking-up score-to-fault table. This method can substantially circumvent those problems of conventional DGA methods by score complementarity, which functions as the score calculated by low-quality data can offset by a high-quality one, and a comprehensive analysis can be done.
Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
