• Proceedings of the KIEE Conference
• /
• 2005.07b
• /
• pp.957-959
• /
• 2005

A noise reduction equipment in power systems was device combined in a passive element such as the form of L, R and C. A analysis of frequency characteristics for the transformer models shall be considered the additional capacitance between the windings($C_{ps}$), terminals($C_s$), and winding-ground($C_g$). A noise in transformer is modeled by take account into the different frequency. In case of the high frequency, transformer windings with stray capacitance have lower impedance. Therefore a noise is transferred from primary winding to secondary winding. Noise Cut Transformer(NCT) was the noise reduction device and was normally single phase two-winding type transformer. We analyzed frequency characteristics using the nodal network analysis with the equivalent circuit and the simulation of the MATLAB.

• 한국전산유체공학회:학술대회논문집
• /
• 2008.03b
• /
• pp.699-702
• /
• 2008

This paper describes the numerical simulations in the cooling of the radiator in a distribution transformer. The aim of this work is the cooling optimization of the transformer by CFD simulations. A clear understanding of the cooling pattern in a radiator which is a main heat remover in the power transformer is essential for optimizing the radiator design increasing the thermal efficiency. In this paper we study the heat transfer and fluid flow in a 3-phase 400kVA transformer. The plate radiators of this transformer become wrinkled (corrugated radiator) and there are filled with transformer oil. The oil is circulated due to the natural convection driven by buoyancy effects through radiators so that the ultimate cooling medium is the surrounding air. In the design of transformers, it is of interest to minimize the cost and size of radiators. The obtained results show the temperature and flow distributions and the possibility to optimize the transformer with 3-dimensional CFD models using FLUENT.

• Proceedings of the KIEE Conference
• /
• 2005.07a
• /
• pp.120-123
• /
• 2005

This paper, for the establishment of power system models and test procedures for the dynamic performance test of transformer protective relay, presents power system model, environment establishement between RTDS and protective relay, and the dynamic performance test generated internal fault or external fault.

• Journal of Power Electronics
• /
• v.11 no.5
• /
• pp.621-631
• /
• 2011

In space qualified DC-DC converters, optimization of the following electrical characteristics is of greater interest in comparison with other specifications; power loss/efficiency, output voltage ripple and volume/weight. The main goal of this paper is to present an appropriate solution for optimizing the above mentioned characteristics. For this purpose, a comprehensive power loss model of a DC-DC converter is fully developed. Proper models are also demonstrated for assessment of the output voltage ripple and the utilized transformer volume as the bulkiest component in a DC-DC converter. In order to provide a test bed for evaluation of the proposed models, a 50W push-pull DC-DC converter is designed and implemented. Finally, a novel cost function with three assigned weight functions is proposed in order to have a trade-off among the power loss, the output voltage ripple and the utilized transformer volume of the converter. The cost function is optimized for applications in which volume has the highest priority in comparison with power loss and ripple. The optimization results show that the transformer volume can be decreased by up to 51% and this result is verified by experimental results. The developed models and algorithms in this paper can be used for other DC-DC converter topologies with some minor modifications.

• Journal of Korea Game Society
• /
• v.19 no.5
• /
• pp.103-114
• /
• 2019

We have been studying a deep-learning program that can communicate with other users in online communities since 2017. But there were problems with processing a Korean data set because of Korean characteristics. Also, low usage of GPUs of RNN models was a problem too. In this study, as Natural Language Processing models are improved, we aim to make better results using these improved models. To archive this, we use a Transformer model which includes Self-Attention mechanism. Also we use MeCab, korean morphological analyzer, to address a problem with processing korean words.

• Journal of the Korean Institute of Illuminating and Electrical Installation Engineers
• /
• v.15 no.6
• /
• pp.58-64
• /
• 2001

This paper is described about design and manufacture of the LVI tester and the model transformer. And we develope the analysis technology of the LVI test with using the developed LVI tester and the model transformer. The LVI tester is consisted of a main body, terminal boxes and cables. It is estimated that the tester has output characteristic to match ANSI specification. The winding deformation models are made of model transformer having tapes and we discuss about the analysis technology of LVI test using the tester and the model transformer.

• Proceedings of the KIEE Conference
• /
• 1991.07a
• /
• pp.281-285
• /
• 1991

In this paper, response characteristics by applying impulse into transformer windings in order to precaution diagonosis of power electric instrument are studied. A sample oilless transformer, an aging transformer and a good transformer are slected for models. The experimented results show that for testing the windings distortion and impedanced change of transformer the LVI method are suitable. Also, variation characteristic of output waveforms at time and frequency domain in order to efficient analysis is discussed. The results response that waveforms at frequency domain is very higher resolution at time domain.

• Journal of the Korea Society of Computer and Information
• /
• v.28 no.9
• /
• pp.17-25
• /
• 2023

In this paper, we propose a model that can perform human pose estimation through a MobileViT-based model with fewer parameters and faster estimation. The based model demonstrates lightweight performance through a structure that combines features of convolutional neural networks with features of Vision Transformer. Transformer, which is a major mechanism in this study, has become more influential as its based models perform better than convolutional neural network-based models in the field of computer vision. Similarly, in the field of human pose estimation, Vision Transformer-based ViTPose maintains the best performance in all human pose estimation benchmarks such as COCO, OCHuman, and MPII. However, because Vision Transformer has a heavy model structure with a large number of parameters and requires a relatively large amount of computation, it costs users a lot to train the model. Accordingly, the based model overcame the insufficient Inductive Bias calculation problem, which requires a large amount of computation by Vision Transformer, with Local Representation through a convolutional neural network structure. Finally, the proposed model obtained a mean average precision of 0.694 on the MS COCO benchmark with 3.28 GFLOPs and 9.72 million parameters, which are 1/5 and 1/9 the number compared to ViTPose, respectively.

• Journal of Power Electronics
• /
• v.9 no.2
• /
• pp.215-223
• /
• 2009

A design methodology for transformers including integrated and center-tapped structures for LLC resonant converters is proposed. In the LLC resonant converter, the resonant inductor in the primary side can be merged in the transformer as a leakage inductance. And, the absence of the secondary filter inductor creates low voltage stress on the secondary rectifiers and is cost-effective. A center-tapped structure of the transformer secondary side is widely used in commercial applications because of its higher efficiency and lower cost than full-bridge structures in the rectifying stages. However, this transformer structure has problems of resonance imbalance and transformer inefficiency caused by leakage inductance imbalance in the secondary side and the position of the air-gap in the transformer, respectively. In this paper, gain curves and soft-switching conditions are derived by first harmonic approximation (FHA) and operating circuit simulation. In addition, the effects of the transformer including integrated and center-tapped structures are analyzed by new FHA models and simulations to obtain an optimal design. Finally, the effects of the air-gap position are analyzed by an electromagnetic field simulator. The proposed analysis and design are verified by experimental results with a 385W LLC resonant converter.

• Transactions on Semiconductor Engineering
• /
• v.1 no.1
• /
• pp.23-30
• /
• 2023

With the recent development of Generative AI technology by IT giants, the size of the transformer model is increasing exponentially over trillion won. In order to continuously enable these AI services, it is essential to reduce the weight of the model. In this paper, we find a hardware-friendly structured pruning pattern and propose a lightweight method of the transformer model. Since compression proceeds by utilizing the characteristics of the model algorithm, the size of the model can be reduced and performance can be maintained as much as possible. Experiments show that the structured pruning proposed when pruning GPT-2 and BERT language models shows almost similar performance to fine-grained pruning even in highly sparse regions. This approach reduces model parameters by 80% and allows hardware acceleration in structured form with 0.003% accuracy loss compared to fine-tuned pruning.