• 대한전기학회:학술대회논문집
• /
• 대한전기학회 2007년도 제38회 하계학술대회
• /
• pp.1106-1107
• /
• 2007

Planar magnetic based design technologies have been widely applied to power design for better cooling and ease of fabrication. The planar transformer and the planar inductor have a low profile characteristics compare to the conventional transformer which would be more cubical in volume. High frequency operation of magnetic components is a main key to achieve high power density of the power module. However, at a high frequency, the skin effect and the proximity effect have to be considered very significantly in magnetic design and also the parasitics in the converter cannot be ignored. This paper deals with the design and the experiment of planar integrated magnetic component. The optimal design for planar magnetics is summarized.

• 전력전자학회논문지
• /
• 제24권4호
• /
• pp.229-236
• /
• 2019

This study proposes a DC-DC converter topology of solid-state transformer for low-voltage DC distribution. The proposed topology consists of a voltage balancer and bipolar DC-DC converter. The voltage and current equations are obtained on the basis of switching states to design the controller. The open-loop gain of the controller is achieved using the derived voltage and current equations. The controller gain is selected through the frequency analysis of the loop gain. The inductance and capacitance are calculated considering the voltage and current ripples. The prototype is fabricated in accordance with the designed system parameters. The proposed topology and designed controller are verified through simulation and experiment.

• 전력전자학회:학술대회논문집
• /
• 전력전자학회 2014년도 전력전자학술대회 논문집
• /
• pp.333-334
• /
• 2014

Converters employing IGCTs usually require di/dt snubber and Over Voltage Protection (OVP) circuit for the protection of IGCTs and fast diodes. In these IGCT-based converters, conventional di/dt snubber and OVP circuit dissipates a significant amount of power loss. To reduce this loss of conventional di/dt snubber and OVP circuit, this paper proposes a flyback type snubber circuit with di/dt limiting characteristic for IGCT-based converters in medium voltage wind turbines. This flyback type snubber circuit simply consists of a flyback type transformer and diode. The proposed circuit reduces loss and simplifies conventional di/dt snubber by adopting the flyback type transformer. Loss analysis of conventional di/dt snubber and OVP circuit is performed for the 3-level NPC type back-to-back VSC supplied from grid voltage of 6.9kV. The proposed flyback type snubber circuit can save the loss of conventional snubber circuit in the 3L-NPC type back-to-back VSC in multi-MW MV wind turbine. The proposed snubber circuit has a fewer number of components and improved efficiency leading to a reliable and efficient wind turbine systems.

• 스마트미디어저널
• /
• 제11권4호
• /
• pp.30-37
• /
• 2022

Disease threatens plant growth and recognizing the type of disease is essential to making a remedy. In recent years, deep learning has witnessed a significant improvement for this task, however, a large volume of labeled images is one of the requirements to get decent performance. But annotated images are difficult and expensive to obtain in the agricultural field. Therefore, designing an efficient and effective strategy is one of the challenges in this area with few labeled data. Transfer learning, assuming taking knowledge from a source domain to a target domain, is borrowed to address this issue and observed comparable results. However, current transfer learning strategies can be regarded as a supervised method as it hypothesizes that there are many labeled images in a source domain. In contrast, unsupervised transfer learning, using only images in a source domain, gives more convenience as collecting images is much easier than annotating. In this paper, we leverage unsupervised transfer learning to perform plant disease recognition, by which we achieve a better performance than supervised transfer learning in many cases. Besides, a vision transformer with a bigger model capacity than convolution is utilized to have a better-pretrained feature space. With the vision transformer-based unsupervised transfer learning, we achieve better results than current works in two datasets. Especially, we obtain 97.3% accuracy with only 30 training images for each class in the Plant Village dataset. We hope that our work can encourage the community to pay attention to vision transformer-based unsupervised transfer learning in the agricultural field when with few labeled images.

• 스마트미디어저널
• /
• 제12권9호
• /
• pp.19-27
• /
• 2023

형태소는 한국어에서 의미를 가진 최소단위이기 때문에, 한국어 언어모델의 성능을 높이기 위해서는 정확한 형태소 분석기의 개발이 필요하다. 기존의 형태소 분석기는 대부분 어절 단위 토큰을 입력 값으로 학습하여 형태소 분석 결과를 제시한다. 하지만 한국어의 어절은 어근에 조사나 접사가 부착된 형태이기 때문에 어근이 같은 어절이어도 조사나 접사로 인해 의미가 달라지는 성향이 있다. 따라서 어절 단위 토큰을 사용하여 형태소를 학습하면 조사나 접사에 대한 오분류가 발생할 수 있다. 본 논문에서는 형태소 단위의 토큰을 사용하여 한국어 문장에 내재된 의미를 과악하고, Transformer를 사용한 시퀀스 생성 방식의 형태소 분석기를 제안한다. 또한, 미등록 단어 문제를 해결하기 위해 학습 말뭉치 데이터를 기반으로 사용자 사전을 구축하였다. 실험 과정에서 각 형태소 분석기가 출력 한 형태소와 품사 태그를 함께 정답 데이터와 비교하여 성능을 측정하였으며, 실험 결과 본 논문에서 제시한 형태소 분석기가 기존 형태소 분석기에 비해 성능이 높음을 증명하였다.

• 스마트미디어저널
• /
• 제12권7호
• /
• pp.43-51
• /
• 2023

온라인 거래가 증가하면서 의류 이미지는 소비자의 구매 결정에 큰 영향을 미치게 되었다. 의류 소재에 대한 이미지 정보의 중요성이 강조되고 있으며, 의류 이미지를 분석하여 사용된 소재를 파악하는 것은 패션 산업에 있어서 중요하다. 의류에 사용된 텍스타일의 소재는 육안으로 식별하기 어렵고, 분류 작업에도 많은 시간과 비용이 소모된다. 본 연구는 딥러닝 알고리즘을 기반으로 의류 이미지로부터 텍스타일의 소재를 분류하고자 하였다. 소재를 분류함으로써 의류 생산 비용을 절감하고, 제조공정의 효율성을 증대하는데 도움이 되며 소비자에게 특정 소재의 제품을 추천하는 AI 서비스에 기여할 수 있다. 의류 이미지를 분류하기 위해 머신비전 기반의 딥러닝 알고리즘 ResNet과 Vision Transformer를 이용하였다. 760,949장의 이미지를 수집하였고, 비정상 이미지를 검출하는 전처리 과정을 거쳤다. 최종적으로 총 167,299장의 의류 이미지와 섬유라벨 19개, 직물라벨 20개를 사용하였다. ResNet과 Vision Transformer를 사용해서 의류 텍스타일의 소재를 분류하였으며 알고리즘 성능을 Top-k Accuracy Score 지표를 통해 비교하였다. 성능을 비교한 결과, ResNet 보다 Vision Transformer 알고리즘이 더 우수하였다.

• International journal of advanced smart convergence
• /
• 제13권2호
• /
• pp.195-204
• /
• 2024

In this paper, we investigate the comparative performance of two leading object detection architectures, YOLOv7 and Detection Transformer (DETR), across varying levels of data augmentation using CycleGAN. Our experiments focus on chest scan images within the context of biomedical informatics, specifically targeting the detection of abnormalities. The study reveals that YOLOv7 consistently outperforms DETR across all levels of augmented data, maintaining better performance even with 75% augmented data. Additionally, YOLOv7 demonstrates significantly faster convergence, requiring approximately 30 epochs compared to DETR's 300 epochs. These findings underscore the superiority of YOLOv7 for object detection tasks, especially in scenarios with limited data and when rapid convergence is essential. Our results provide valuable insights for researchers and practitioners in the field of computer vision, highlighting the effectiveness of YOLOv7 and the importance of data augmentation in improving model performance and efficiency.

• 전력전자학회논문지
• /
• 제22권4호
• /
• pp.336-344
• /
• 2017

This paper presents the three-port DC-DC converter modeling and controller design procedure, which is part of the solid-state transformer (SST) to interface medium voltage AC grid to bipolar DC distribution network. Due to the high primary side DC link voltage, the proposed converter employs the three-level neutral point clamped (NPC) topology at the primary side and 2-two level half bridge circuits for each DC distribution network. For the proposed converter particular structure, this paper conducts modeling the three winding transformer and the power transfer between each port. A decoupling method is adopted to simplify the power transfer model. The voltage controller design procedure is presented. In addition, the output current sharing controller is employed for current balancing between the parallel-connected secondary output ports. The proposed circuit and controller performance are verified by experimental results using a 30 kW prototype SST system.

• Journal of Power Electronics
• /
• 제14권3호
• /
• pp.421-431
• /
• 2014

A DSP-based self-adaptive proportional-integral (PI) controller to control a DC-DC converter is proposed in this paper. The full-bridge topology is adopted here to obtain higher power output capability and higher conversion efficiency. The converter adopts the zero-voltage-switching (ZVS) technique to reduce the conduction losses. A parallel secondary active clamp circuit is added to deal with the voltage overshoot and ringing effect on the transformer's secondary side. A self-adaptive PI controller is proposed to replace the traditional PI controller. Moreover, the designed converter adopts the constant-current and constant-voltage (CC-CV) output control strategy. The secondary active clamp mechanism is discussed in detail. The effectiveness of the proposed converter was experimentally verified by an IGBT-based 10kW prototype.

• 한국자기학회:학술대회 개요집
• /
• 한국자기학회 2012년도 임시총회 및 하계학술연구발표회
• /
• pp.136-136
• /
• 2012