A Comparative Study of Neural Network Efficiency in Power Transformers Diagnosis Using Dissolved Gas Analysis

TLDR

The study compares neural network efficiency for detecting early faults in power transformers. The NN was trained on five dissolved‑gas‑analysis criteria (Doemenburg, modified Rogers, Rogers, IEC, CSUS), tested on new data, and its diagnoses were compared with inspection results. The NN achieved 87–100 % successful diagnosis depending on the criterion.

Abstract

This paper presents a comparative study of neural network (NN) efficiency for the detection of incipient faults in power transformers. The NN was trained according to five diagnosis criteria commonly used for dissolved gas analysis (DGA) in transformer insulating oil. These criteria are Doemenburg, modified Rogers, Rogers, IEC, and CSUS. Once trained, the NN was tested by using a new set of DGA results. Finally, NN diagnosis results were compared with those obtained by inspection and analysis. The study shows that the NN rate of successful diagnosis is dependant on the criterion under consideration, with values in the range of 87-100%.