• 대한설비공학회:학술대회논문집
• /
• 대한설비공학회 2009년도 하계학술발표대회 논문집
• /
• pp.955-960
• /
• 2009

In order to improve the efficiency of the main transformer in a tilting train, the optimal operation of a cooling system is necessary. Mathematical models of a main transformer cooling system were developed. These include models for the main transformer, the oil pump, the oil cooler, and the blower. The optimal oil temperature algorithm was also developed. This consists of the optimal setpoint algorithm and the control algorithm. A simulation program was developed by using mathematical models and the optimal oil temperature algorithm. Simulation results showed that the dynamic behavior of a main transformer cooling system was predicted well by mathematical models and a main transformer cooling system was controlled effectively by the optimal oil temperature algorithm.

• 전기학회논문지
• /
• 제58권10호
• /
• pp.2000-2004
• /
• 2009

In general, for the condition monitoring of a power transformer using the UHF PD measuring technique, detection of any partial discharge, identifying the defect in the transformer and locating the insulation defect are necessary. In this paper one of the most frequent detects which can result in turn to turn fault in power transformer was examined for identifying the defect. In order to model the defect, as a discharge source, a partial discharge cell was used for experimental activity. Magnitude of electromagnetic wave signals and corresponding amount of apparent discharge were measured simultaneously against phase of applied voltage to the discharge cell. Frequency range and phase resolved partial discharge signals were measured and analyzed. The results will be contributed to build the defect database of power transformer and to decrease the occurrence of transformer faults.

• 한국정보처리학회:학술대회논문집
• /
• 한국정보처리학회 2022년도 춘계학술발표대회
• /
• pp.546-548
• /
• 2022

최근 컴퓨터 비전 분야에서 Transformer를 도입한 연구가 활발히 연구되고 있다. 이 모델들은 Transformer의 구조를 거의 그대로 사용하기 때문에 확장성이 좋으며 large 스케일 학습에서 매우 우수한 성능을 보여주었다. 하지만 Transformer를 적용한 비전 모델은 inductive bias의 부족으로 학습 시 많은 데이터와 시간을 필요로 하였다. 그로 인하여 현재 많은 Vision Transformer 개선 모델들이 연구되고 있다. 본 논문에서도 Vision Transformer의 문제점을 개선한 Clustering CoaT 모델을 제안한다.

• Journal of Electrical Engineering and Technology
• /
• 제13권2호
• /
• pp.781-789
• /
• 2018

This paper proposes a new approach utilizing empirical mode decomposition (EMD) reconstruction to process vibration signals of a transformer under DC bias caused by high voltage direction transmission (HVDC), which is the potential cause of additional vibration and noise from transformer. Firstly, the Calculation Method is presented and a 3D model of transformer is simulated to analyze transformer deformation characteristic and the result indicate the main vibration is produced along axial direction of three core limbs. Vibration test system has been built and test points on the core and shell of transformer have been measured. Then, the signal reconstruction method for transformer vibration based on EMD is proposed. Through the EMD decomposition, the corrupted noise can be selectively reconstructed by the certain frequency IMFs and better vibration signals of transformer have been obtained. After EMD reconstruction, the vibrations are compared between transformer in normal work and with DC bias. When DC bias occurs, odd harmonics, vibration of core and shell, behave as a nonlinear increase and the even harmonics keep unchanged with DC current. Experiment results are provided to collaborate our theoretical analysis and to illustrate the effectiveness of the proposed EMD method.

• 한국수자원학회논문집
• /
• 제57권7호
• /
• pp.437-449
• /
• 2024

딥러닝을 활용하여 유역 특성을 반영한 유량 예측 및 비교 연구가 주목받고 있다. 본 연구는 셀프 어텐션 메커니즘을 통해 대용량 데이터 훈련에 적합한 Transformer와 인코더-디코더(Encoder-Decoder) 구조를 가지는 LSTM-based multi-state-vector sequence-to-sequence (LSTM-MSV-S2S) 모형을 선정하여 유역정보(catchment attributes)를 고려할 수 있는 모형을 구축하였고 이를 토대로 국내 10개 다목적댐 유역의 유입량을 예측하였다. 본 연구에서 설계한 실험 구성은 단일유역-단일훈련(Single-basin Training, ST), 다수유역-단일훈련(Pretraining, PT), 사전학습-파인튜닝(Pretraining-Finetuning, PT-FT)의 세 가지 훈련 방법을 사용하였다. 모형의 입력 자료는 선정된 10가지 유역정보와 함께 기상 자료를 사용하였으며, 훈련 방법에 따른 유입량 예측 성능을 비교하였다. 그 결과, Transformer 모형은 PT와 PT-FT 방법에서 LSTM-MSV-S2S보다 우수한 성능을 보였으며, 특히 PT-FT 기법 적용 시 가장 높은 성능을 나타냈다. LSTM-MSV-S2S는 ST 방법에서는 Transformer보다 높은 성능을 보였으나, PT 및 PT-FT 방법에서는 낮은 성능을 보였다. 또한, 임베딩 레이어 활성화 값과 원본 유역정보를 군집화하여 모형의 유역 간 유사성 학습 여부를 분석하였다. Transformer는 활성화 벡터가 유사한 유역들에서 성능이 향상되었으며, 이는 사전에 학습된 다른 유역의 정보를 활용해 성능이 개선됨을 입증하였다. 본 연구는 다목적댐별 적합한 모형 및 훈련 방법을 비교하고, 국내 유역에 PT 및 PT-FT 방법을 적용한 딥러닝 모형 구축의 필요성을 제시하였다. 또한, PT 및 PT-FT 방법 적용 시 Transformer가 LSTM-MSV-S2S보다 성능이 더 우수하였다.

• 한국조명전기설비학회:학술대회논문집
• /
• 한국조명전기설비학회 1990년도 추계학술발표회논문집
• /
• pp.53-56
• /
• 1990

When oil flows and rubs against various materials in transformer, electrostatic charges are separated at the interface of the oil and the solid material. Using simplified model transformer, authors investigated the basic characteristics of the streaming electrifica-tion which is caused by forced oil circulation. As the result of the study, it was concluded that the electrostatic charge distribution on test pipe of the transformer showed larger leakage current at the inlet and the outlet.

• Journal of Magnetics
• /
• 제19권3호
• /
• pp.295-299
• /
• 2014

Recently, module-integrated converter (MIC) research has shown interest in small-scale photovoltaic (PV) generation. The converter is capable of efficient power generation. In this system, the high frequency transformer should be made compact, and demonstrate high efficiency characteristics. This paper presents a core structure optimization procedure to improve the efficiency of a high frequency transformer of compact size. The converter circuit is considered in the finite element analysis (FEA) model, in order to obtain an accurate FEA result. The results are verified by the testing of prototypes.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 1999년도 하계학술대회 논문집 A
• /
• pp.86-88
• /
• 1999

An analysis of temperature distribution in transformer is necessary for cooling design. But, it is very difficult to make that analysis because of the complicated structure of transformer. Particulary. if it is asymmetry, 3 dimensional analysis is required. This paper presents the 3-dimensional analysis technique of temperature distribution in transformer using a commercial CFD program FLUENT and the applied results in a simple model.

• 한국전기전자재료학회:학술대회논문집
• /
• 한국전기전자재료학회 2003년도 하계학술대회 논문집 Vol.4 No.1
• /
• pp.212-215
• /
• 2003

In this paper, a model pole transformer was designed to improve the reliability of pole transformers and its short circuit withstand strength was tested. The failure causes of the pole transformers which had been operated in the field were analyzed and several patterns of the causes were determined. A novel countermeasure was devised to prevent the failures of the pole transformers in advance. A model pole transformer was designed and made in accordance with the given specifications to strengthen the short circuit withstand capability and its characteristic was tested. As a result, the advanced performance of the model pole transformer was confirmed and novel techniques in manufacturing process was suggested. It is considered that all the methods employed in the developing process will be helpful to design and manufacture the pole transformers thus the reliability of pole transformers in distribution lines will be improved.

• 대한임베디드공학회논문지
• /
• 제18권5호
• /
• pp.217-223
• /
• 2023

In this paper, we propose a new model for the conditional generation of music, considering key and rhythm, fundamental elements of music. MIDI sheet music is converted into a WAV format, which is then transformed into a Mel Spectrogram using the Short-Time Fourier Transform (STFT). Using this information, key and rhythm details are classified by passing through two Convolutional Neural Networks (CNNs), and this information is again fed into the Music Transformer. The key and rhythm details are combined by differentially multiplying the weights and the embedding vectors of the MIDI events. Several experiments are conducted, including a process for determining the optimal weights. This research represents a new effort to integrate essential elements into music generation and explains the detailed structure and operating principles of the model, verifying its effects and potentials through experiments. In this study, the accuracy for rhythm classification reached 94.7%, the accuracy for key classification reached 92.1%, and the Negative Likelihood based on the weights of the embedding vector resulted in 3.01.