• Proceedings of the KIEE Conference
• /
• 1995.11a
• /
• pp.21-23
• /
• 1995

The analysis of the thin lamination model of the wound core type transformer by FEM requires many region division and much calculationg time, and it has difficulty in calculation by modem computer. A new method which is simulated by a solid but anisotropic block with the magnetic permeabilities in the orthogonal dimension, is selected to account for the presence of the lamination.[1] Based on this equivalent anisotropic block model, we analyze the iron loss of the wound core type transformer by FEM, and compare the simulation results with the experimental results.

• Proceedings of the KIEE Conference
• /
• 1994.07a
• /
• pp.156-158
• /
• 1994

The analysis of the thin lamination model of the amorphous transformer by FEM requires many region, divisions and much calculating tine, and it has difficulty in calculating for mordern computer. A new method which is simulated by a solid but anisotropic block with the magnetic permeabilities in two orthogonal directions selected to account for the presence of the laminations. [1] Based on this equivalent anisotropic block model, we analyzed the iron loss of the amorphous transformer by FEM, and presented an optimal design example of core dimensions for minimizing the iron loss.

• The Proceedings of the Korean Institute of Illuminating and Electrical Installation Engineers
• /
• v.7 no.1
• /
• pp.37-45
• /
• 1993

A parametric transformer, as static power converter operating on the principle of parametric excitation, is analysed. For the purpose of quantitative analysis of device, the mathematical model of the device is derived. On the basis of this model, the performances of the parametric transformer, such as over and under voltage protection, overload protection, bilateral filtering and frequency multiplication, are obtained quantitatively and analysed qualitatively.

• Proceedings of the Korean Institute of Electrical and Electronic Material Engineers Conference
• /
• 2000.07a
• /
• pp.423-426
• /
• 2000

The 3-D $31.5$\times$31.5$\times$2.5mm$^3$model of contour vibration mode piezoelectric transformer with Pb($Ni_{1/2}$$W_{1/2}$)$O_3$-Pb(Zr,Ti)$O_3$ ceramics was simulated by ANSYS according to the dot size 17, 18, 19, 20, 21 $mm\phi$ and analyzed the results. The mechanical quality factor of the 3-D model was decreased with the dot size and increased as 2605 at 20 $mm\phi$ and after then decreased again. The output voltage of that sample was 74123 [V] and the maximum stress of the dot electrode at that sample was 288[$10^7$N/$m^2$] and the maximum displacement of the ring electrode at that sample was 128 [$\mu\textrm{m}$].

• Annual Conference of KIPS
• /
• 2020.05a
• /
• pp.523-525
• /
• 2020

In recent years, deep learning has been widely used in news recommendation systems. In the previous personalized news recommendation, a large number of CF-based methods, content-based or hybrid methods have been proposed. But most of the works are only modeling the user's interaction history, ignoring the hidden meaning of the user's continuous behaviors. In this paper, we propose to adopt the powerful Transformer model in order to understand the hidden meaning of the user's continuous behaviors in news recommendations. The experimental results prove the superiority of the transformer, and the AUC has been significantly improved as compared to the original model.

• Journal of Information Processing Systems
• /
• v.20 no.4
• /
• pp.458-476
• /
• 2024

To address the issue of class imbalance in network traffic data, which affects the network intrusion detection performance, a combined framework using transformers is proposed. First, Tomek Links, SMOTE, and WGAN are used to preprocess the data to solve the class-imbalance problem. Second, the transformer is used to encode traffic data to extract the correlation between network traffic. Finally, a hybrid deep learning network model combining a bidirectional gated current unit and deep neural network is proposed, which is used to extract long-dependence features. A DNN is used to extract deep level features, and softmax is used to complete classification. Experiments were conducted on the NSLKDD, UNSWNB15, and CICIDS2017 datasets, and the detection accuracy rates of the proposed model were 99.72%, 84.86%, and 99.89% on three datasets, respectively. Compared with other relatively new deep-learning network models, it effectively improved the intrusion detection performance, thereby improving the communication security of network data.

• The Transactions of the Korean Institute of Power Electronics
• /
• v.22 no.2
• /
• pp.102-108
• /
• 2017

Proper design guides are proposed for a practical dual-active bridge (DAB) converter based on the mathematical model on the steady state. The DAB converter is popular in bidirectional application due to its zero-voltage capability and easy bidirectional operation for seamless control, high efficiency, and performance. Some design considerations are taken to overcome the limitation of the DAB converter. The practical design methodology of power stage is discussed to minimize the conduction and switching losses of the DAB converter. Small-signal model and frequency response are derived and analyzed based on the generalized average method, which considers equivalent series resistance, to improve the dynamics, stability, and reliability with voltage regulation of the practical DAB converter. The design of closed-loop control is discussed by the derived small-signal model to obtain the pertinent gain and phase margin in steady-state operation. Experimental results of a 3.3 kW prototype of DAB converter demonstrate the validity and effectiveness of the proposed methods.

• The Journal of Information Systems
• /
• v.32 no.4
• /
• pp.149-162
• /
• 2023

Purpose In the textile industry, fabric defects significantly impact product quality and consumer satisfaction. This research seeks to enhance defect detection by developing a transformer-based deep learning image segmentation model for learning high-dimensional image features, overcoming the limitations of traditional image classification methods. Design/methodology/approach This study utilizes the ZJU-Leaper dataset to develop a model for detecting defects in fabrics. The ZJU-Leaper dataset includes defects such as presses, stains, warps, and scratches across various fabric patterns. The dataset was built using the defect labeling and image files from ZJU-Leaper, and experiments were conducted with deep learning image segmentation models including Deeplabv3, SegformerB0, SegformerB1, and Dinov2. Findings The experimental results of this study indicate that the SegformerB1 model achieved the highest performance with an mIOU of 83.61% and a Pixel F1 Score of 81.84%. The SegformerB1 model excelled in sensitivity for detecting fabric defect areas compared to other models. Detailed analysis of its inferences showed accurate predictions of diverse defects, such as stains and fine scratches, within intricated fabric designs.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.66 no.9
• /
• pp.1334-1340
• /
• 2017

In this paper, a tie-plate shape is optimized by using the numerical technique to reduce the stray load loss of the tie-plate which is a mechanical structure for assembling and supporting of the transformer core. The eddy current loss of the structure is calculated by an electromagnetic field FEM program and the results are compared with 4 different shapes of tie-plates. Since the thickness of the tie-plate is very thin, and the skin depth is very small, the number of FE elements for 3-D transformer model is too big to solve. So, the surface impedance boundary condition (SIBC) is used to reduce the system matrix size and its computing time. To verify the method a 2.5 MVA 22,900/380V distribution transformer is simulated using one objective function and three design variables with some constraints. The final optimized tie-plate has three slots of 6 mm width and 23 mm gap, and the loss is reduced by 75 %. Consequently, the proposed algorithm seems to be considerably applicable to electric machinery as well as power transformer.

• The Transactions of the Korean Institute of Electrical Engineers B
• /
• v.48 no.7
• /
• pp.376-383
• /
• 1999

The piezoelectric transformer (PT) is an electro-mechanical device that transfers electrical energy through a mechanical vibration. In this paper, a PT operating in the contour vibration mode is introduced for an application of fluorescent lamp ballast. Utilizing its inherent characteristics of the LC resonator and a high voltage gain to ignite the lamp in light load condition, an investigation of a power piezoelectric transformer as a potential component for a fluorescent lamp ballast is discussed. PT is easy to be produced in mass and reduces the cost of the ballast. The modified equivalent circuit model of the PT considering the operating current level is derived to design the fluorescent lamp ballast. This model describes the voltage gain of the PT in wide load variations and various input current levels. The experimental and simulation results are provided to verify theoretical analysis. The power capacity of the currently developed PT is relatively low (15W), but it can be increased by adopting a multi-layer structure and is currently under investigation. It is also possible to parallel the PT for higher power processing.