• KIISE Transactions on Computing Practices
• /
• v.23 no.9
• /
• pp.542-549
• /
• 2017

Sequence-to-sequence models and similar pointer networks suffer from performance degradation when an input is composed of multiple sentences or when the length of the input sentence is long. To solve this problem, this paper proposes a hierarchical pointer network model that uses both the word level and sentence level information to encode input sequences composed of several sentences at the word level and sentence level. We propose a hierarchical pointer network based coreference resolution that performs a coreference resolution for all mentions. The experimental results show that the proposed model has a precision of 87.07%, recall of 65.39% and CoNLL F1 74.61%, which is an improvement of 21.83% compared to an existing rule-based model.

• Proceedings of the KIEE Conference
• /
• 2005.07a
• /
• pp.371-373
• /
• 2005

An enhancement for a transmission and substation equipment in power system make the system impedance to be lower. In principle, if the system impedance become low, system stability will be better, but the fault current become very higher. It is a very big problem for CB operating. As a fact of CB operating performance, high amplitude of the fault current may cause CB operation failure because of exceeding standard value in TRV. So we simulated TRV by using the EMTP. Generally there are two types of TRV in actual power system. One is short line fault, the other is bus terminal fault. In this paper, we simulated the TRv at short line fault as installed current limit reactors to reduce fault current in 345kV ultra-high voltage system. Short line fault is caused from single line fault in transmission line.

• Journal of the Korean Institute of Telematics and Electronics
• /
• v.26 no.1
• /
• pp.88-97
• /
• 1989

A new system for three dimensional object recognition is proposed in an attempt to overcome the lack of accuracy and flexibility, the essential problem at which model-based recognition technique faces. Hierarchical data base was designed to manage a number of features efficiently which are extracted for the purpose of matching by the system and to provide information necessary to match systema6tically. The results of matching, in turn, is fed back to the data base, and stored in. The recognition is done by saving the results on higher level of hierachical data base.

• Journal of KIISE
• /
• v.45 no.1
• /
• pp.45-52
• /
• 2018

With the advent of BCI technology in recent years, various BCI products have been released. BCI technology enables brain information to be transmitted directly to a computer, and it will bring a lot of convenience to life. However, there is a problem with information protection. In particular, EEG data can raise issues about personal privacy. Collecting and analyzing big data on EEG reports raises serious concerns about personal information exposure. In this paper, we propose a secure privacy-preserving BCI model in a big data environment. The proposed model could prevent personal identification and protect EEG data in the cloud environment.

• Journal of the Korean Society for Precision Engineering
• /
• v.28 no.10
• /
• pp.1141-1145
• /
• 2011

Recently, there are many issues about R2R printing technique for mass production of electronic devices. Among the various Roll-to-roll based printing techniques such as gravure, off-set, flexo and so on, "Gravure off-set printing technique" has an advantage of higher printing resolution. The printing unit of gravure off-set printing technique usually consists of plate roll, blanket roll and impressure roll whose. Linear velocities should be synchronized each other for fine pattern printing. However, roller's manufacturing error and printing variations such as pringting pressure, printing speed, roll stroke and so on actually affected their synchronization anf thus the quality of fine fattern. In this paper, we analyzed the effective of synchronization error on printing quality. Also, this paper reviews the relative motion with each roll. And, this paper studys the synchronization error about its generation problem.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.12 no.2
• /
• pp.361-366
• /
• 2017

The abnormal state of the main equipment across the industry is accompanied by abnormal rise of temperature, change in the vibration and noise. In this study, we developed a smart sensing module equipped with BLE, and developed a smart sensing device that can detect abnormal vibration due to its own flaws of the equipment by interfacing with an MEMS-based acceleration sensor. The smart sensing device developed in this study can be easily installed in a small space and can monitor the vibration status of the equipment in real time, and can easily inform the user of the steady state and the problem occurrence status with array LED display.

• The Transactions of the Korean Institute of Power Electronics
• /
• v.18 no.5
• /
• pp.477-484
• /
• 2013

This paper proposes a novel switching method of active clamp snubber for efficiency improvement of PV module integrated converter(MIC) system. Recently, MIC solar system is researched about the efficiency and safety. PV MIC system is used active clamp method of snubber circuit for the price and reliability of the system. But active clamp snubber circuit has the disadvantage that system efficiency is decreased for switch operating time because of heat loss of resonant between snubber capacitor and leakage inductance. To solve this problem, this paper proposes a novel switching method of the active clamp. The proposed method is a technique to reduce power consumption by reducing the resonance of the snubber switch operation time and through simulations and experiments proved the validity.

• The Transactions of the Korean Institute of Power Electronics
• /
• v.25 no.1
• /
• pp.54-60
• /
• 2020

A digital current control scheme using virtual d-q transformation for a boost single-phase power factor correction (PFC) converter is proposed. The use of virtual d-q transformation in single-phase power converters is known to improve current control performance. However, the conventional virtual d-q transformation-based digital current control scheme cannot be directly applied to the boost single-phase PFC converter because the current and average voltage waveforms of the inductor used in the converter are not sinusoidal. To cope with this problem, this study proposes a virtual sinusoidal signal generation method that converts the current and average voltage waveform of the inductor into a sinusoidal waveform synchronized with the grid. Simulation and experimental results are provided to show that the virtual d-q transformation-based digital current control is successfully applied to the boost single-phase PFC converter with the aid of the proposed virtual sinusoidal signal generation method.

• The Transactions of the Korean Institute of Power Electronics
• /
• v.22 no.2
• /
• pp.95-101
• /
• 2017

A simple and practical voltage balance method for a solid-state transformer (SST) is proposed to reduce the voltage difference of cascaded H-bridge converters. The tolerance device components in SST cause the imbalance problem of DC-link voltage in the H-bridge converter. The Max/Min algorithms of voltage balance controller are merged in the controller of an AC/DC rectifier to reduce the voltage difference. The DC-link voltage through each H-bridge converter can be balanced with the proposed control methods. The design and performance of the proposed SST are verified by experimental results using a 30 kW prototype.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2014.10a
• /
• pp.721-724
• /
• 2014

Ultrasonic sensor is used widely for many applications because low cost, simple structure, and low restriction. There are many difficulties to recognize an object by use an ultrasonic sensor, because of low resolution, poor direction, and measurement error. To improve the these problem, we use the various kinds of sensor arrangement methods, large amount of sensor, and change the arrangement pattern of sensor. In this paper, to obtain the most basic parameters for pattern recognition such as distance, dimension of the object, an angle of the object, we get the improved results by use the intelligent calculation algorithm based on Neuro-Fuzzy. This method use the multifarious output voltage of ultrasonic sensor by simple electronic circuit.