• 예술인문사회 융합 멀티미디어 논문지
• /
• 제8권3호
• /
• pp.563-570
• /
• 2018

본 논문에서는 기계학습 기반의 효율적인 변압기 유중가스 분석 및 분류 방법을 제안하여 기존 IEC 60599 진단기준 기반의 문제점을 해결하고 진단 성능을 개선한다. 기존 IEC 60599 진단기준은 조성비가 진단 기준에 존재하지 않거나 경계조건에 있는 경우 진단 전문가에게 의뢰하지 않고는 해석에 어려움이 있으며 진단영역이 겹치는 부분이 존재하므로 정확한 원인분석을 수행하는 데에 한계가 있다. 따라서 IEC 60599 진단 기준만으로 변압기 유중가스 데이터를 분석 및 분류하는 경우 IEC 60599 기준에 만족하지 않는 데이터를 분류하지 못한다는 문제점이 있다. 이와 같은 문제를 해결하기 위해 기계학습 기반의 변압기 유중가스 분석 및 분류 방법을 제안하였다. 제안한 기계학습 기반의 변압기 유중가스 분석 방법은 IEC 60599 진단기준으로 판단이 불가능한 데이터를 서포트 벡터 머신을 통해 정확히 분류 할 수 있다. 제안한 방법의 성능을 검증하기 위해 실제 유중가스 데이터를 사용하여 기계학습 기반의 변압기 유중가스 분석 방법의 효율성을 검증하였다.

• 전기학회논문지P
• /
• 제65권1호
• /
• pp.41-46
• /
• 2016

It has been proven that the dissolved gas analysis (DGA) is the most effective and convenient method to diagnose the transformers. The DGA is a simple, inexpensive, and non intrusive technique. Among the various diagnosis methods, IEC 60599 has been widely used in transformer in service. But this method cannot offer accurate diagnosis for all the faults. This paper proposes a fault diagnosis method of oil-filled power transformers using IEC code based neuro-fuzzy model. The proposed method proceeds two steps. First, IEC 60599 method is applied to diagnosis. If IEC code can't determine the fault type, neuro-fuzzy model is applied to effectively classify the fault type. To demonstrate the validity of the proposed method, experiment is performed and its results are illustrated.

• 한국조명전기설비학회:학술대회논문집
• /
• 한국조명전기설비학회 2007년도 추계학술대회 논문집
• /
• pp.275-278
• /
• 2007

Dissolved gases analysis (DGA) is one of the most useful techniques to detect incipient faults in power transformers. Criteria interpreting DGA result is the most important. Because of difference of operation environment, construction type, oil volume, and etc, the interpretative criteria of DGA at KEPCO must be different with other standard like IEC-60599, Rogers and Doernenburg. In this paper, we collected the DGA data of the normal condition transformers and the incipient fault transformer to suggest the most appropriate criteria. Using these data, this paper suggests appropriate condition classification algorithm. Suggested algorithm can help to detect incipient fault earlier without unnecessary sampling.

• 전기학회논문지P
• /
• 제65권3호
• /
• pp.188-193
• /
• 2016

It has been proven that the dissolved gas analysis (DGA) is the most effective and convenient method to diagnose the transformers. The DGA is a simple, inexpensive, and non intrusive technique. Among the various diagnosis methods, IEC 60599 has been widely used in transformer in service. But this method cannot offer accurate diagnosis for all the faults. This paper proposes a fault diagnosis method of oil-filled power transformers using DGA and Intelligent Probability Model. To demonstrate the validity of the proposed method, experiment is performed and its results are illustrated.

• 조명전기설비학회논문지
• /
• 제21권9호
• /
• pp.87-94
• /
• 2007

This paper presents new criteria for DGA(Dissolved Gases Analysis) using CDF(Cumulative Distribution Function) obtained from the data from the diagnosis of transformers operated in KEPCO over a period of 16 years. Because of differences in operating environments, construction type, oil volume, and other factors, the interpretative criteria of DGA at KEPCO differs from other standards such as IEC-60599, or Rogers and Doernenburg. To suggest the most appropriate criteria, the DGA data from transformers under normal conditions as well as from developing fault transformers were collected. Using these data, this study suggests the limitative gas level of transformers under normal operating conditions and verifies the suitability of the criteria. Because the application of this new criterion to transformers at KEPCO increases the detectable ratio of incipient faults and reduces unnecessary follow-up sampling and analysis, the new criteria yields a more reliable prediction of transformer condition.