• 문화기술의 융합
• /
• 제7권1호
• /
• pp.668-673
• /
• 2021

4차산업혁명이 본격화됨에 따라 관련 기술들이 화두가 되고 있다. 고속 무선통신과 같은 기술을 최대한으로 활용하기 위한 하드웨어 개발이 가속화되고 있으며, 관련 기업들이 급격히 성장하고 있다. 인공지능의 경우 관련 연구를 위해서 일반적으로 데스크톱을 사용하는 경우가 많지만, 주로 딥러닝의 학습 과정을 위해 사용되고 있으며 생성된 모델을 프로그램 등에 포함하여 사용할 기기에 이식하는 경우가 많다. 하지만, 학습량이 과도하거나 데스크톱의 성능만큼 제작된 모델을 사용하게 되어 전원공급이 따로 이루어지지 않는 기기의 경우 전력이 부족하거나 성능이 충분하지 못하기 때문에 제 결과를 내기 어렵다. 본 논문에서는 딥러닝의 성능을 임베디드 보드에 맞추어 개발하기 전에 판매되고 있는 몇 가지 Neural Process Unit을 탑재한 보드와 USB로 딥러닝 수행 성능을 높일 수 있는 딥러닝 액셀러레이터를 사용하여 효율성을 비교하여 임베디드 보드로 가능한 개발 방향을 제시한다.

• 한국전기전자재료학회:학술대회논문집
• /
• 한국전기전자재료학회 2008년도 추계학술대회 논문집 Vol.21
• /
• pp.264-265
• /
• 2008

While the reliability of bulk insulation has become important particularly in multilayer boards and embedded boards, electronics are to be used under various environments such as at high temperature and in high humidity. We observed internal space charge behavior for two types of epoxy composites under dc electric fields to investigate the influence of water at high temperature. In the case of glass/epoxy specimen, homocharge is observed at water-treated specimen, and spatial oscillations become clearer in the water-treated specimens. Electric field in the vicinity of the electrodes shows the injection of homocharge. In aramid/epoxy specimens, heterocharge is observed at water-treated specimens, i.e. negative charge accumulates near the anode, while positive charge accumulates near the cathode. Electric field is enhanced just before each electrode. In order to further examine the mechanism of space charge formation, we have developed a new system that allows in situ space charge observation during ion migration tests at high temperature and high humidity. Using this in situ system.

• 한국소프트웨어감정평가학회 논문지
• /
• 제17권2호
• /
• pp.161-172
• /
• 2021

뉴로모픽 아키텍처는 저에너지로 인공지능 기술을 지원하는 차세대 컴퓨팅으로 주목받고 있다. 그러나 뉴로모픽 아키텍처 기반의 FPGA 임베디드 보드는 크기나 전력 등으로 인하여 가용 자원이 제한된다. 본 논문에서는 제한된 자원을 효율적으로 사용하기 위해 특징점의 고려 없이 크기를 재조정하는 보간법과 에너지 기반으로 특징점을 최대한 보존하는 DCT(Discrete Cosine Transform) 기법을 통한 특징 표현 방법을 비교 및 평가한다. 크기가 조정된 이미지는 일반적인 PC 환경에서와 FPGA 임베디드 보드의 Nengo 프레임워크에서 컨벌루션 신경망을 통해 정확도를 비교 분석했다. 실험 결과 PC의 컨벌루션 신경망과 FPGA Nengo 환경 모두에서 DCT 기반 분류 성능이 일반 보간법보다 약 1.9% 높은 성능을 보였다. 실험 결과를 바탕으로 뉴로모픽 구조 기반 FPGA 보드의 제한된 자원 환경에서 기존에 사용되던 보간법 대신 DCT 방식을 이용한다면 분류에 사용되는 뉴런의 표현에 많은 자원을 할당하여 인식률을 높일 수 있을 것으로 기대한다.

• 한국전기전자재료학회:학술대회논문집
• /
• 한국전기전자재료학회 2007년도 하계학술대회 논문집 Vol.8
• /
• pp.232-233
• /
• 2007

We manufactured Liquid Crystal Polymer (LCP) and (1-x)CaTiO3-xLaAlO3 (CT-LA) ceramic composites and investigated dielectric properties to use as embedded capacitor in printed circuit boards and replace LTCC substrate. The dielectric properties of these composites are varied with volume fraction of CT-LA and ratios of CT/LA. Dielectric constants are in the range of 3~15. In addition, we could get low TCC and High Q value that could not achieve in other ceramic-polymer composites. Especially, in composite with x=0.01 and 30 vol% CT-LA, the dieletric constant and Q-value are 10 and 200, respectively. And more TCC is $-28{\sim}300ppm/^{\circ}C$ in the temperature range of $-55{\sim}125^{\circ}C$. We think that this composites can be used high-Q substrate material like LTCC and embedded temperature compensation capacitor in printed circuit boards.

• 한국정보통신학회:학술대회논문집
• /
• 한국정보통신학회 2015년도 추계학술대회
• /
• pp.651-653
• /
• 2015

Arduino, Raspberry Pi, BeagleBone Black 등 임베디드 보드가 세상에 보편화 되어 임베디드 보드를 활용하여 홈 스마트 시스템, 공작 기계 등 여러 제품들이 만들어 진다. 최근 시약이나 위험물등 위험한 재료를 관리하여 다루는데 있어 사고 및 화재 발생 시 미리 알지 못해 방치되어 큰 피해가 빈번히 일어나고 있다. 이에 본 논문에서는 임베디드(BeagleBone Black)보드, 센서, 그리고 RFID를 활용 하여 실시간으로 위험 재료들을 보관하여 관리하는 시스템을 제안한다. 이는 문제 발생 시 관리자에게 즉시 정보를 제공하여 안전사고를 미연에 예방 할 수 있을 것으로 판단된다.

• 한국전기전자재료학회:학술대회논문집
• /
• 한국전기전자재료학회 2008년도 하계학술대회 논문집 Vol.9
• /
• pp.60-60
• /
• 2008

We manufactured Liquid Crystal Polymer (LCP) and (1-x)$BaTiO_3-xSrTiO_3$(BST) ceramic composites and investigated dielectric properties to use as embedded capacitor in printed circuit boards and replace LTCC substrates. The dielectric properties of these composites are varied with volume fraction of BST and ratios of BT/ST. Dielectric constants are in the range of 3~28. In addition, we could get low TCC and High Q value that could not achieve in other ceramic-polymer composites. Especially, in composite with x=0.4 and 50vol% BST, the dieletric constant and Q-value are 27 and 300, respectively. And more TCC is -116~145ppm/$^{\circ}C$ in the temperature range of -55~$125^{\circ}C$. We think that this composites can be used high-Q substrate material like LTCC and embedded temperature compensation capacitor in printed circuit boards.

• Journal of Advanced Marine Engineering and Technology
• /
• 제28권5호
• /
• pp.811-817
• /
• 2004

Recently embedded systems are widely used in various industrial fields as supervisory controller because they have many merits. One of merits seems to be that operating environment of embedded system is the same as development environment using PC. That makes developing and manufacturing period shorten and also proper time to market Most of all machinery have sequential control system for their maneuvering which is composed of relays. contacts. timers. etc. In this paper. software sequential control system is proposed to be able to replace hardware sequential control system by using embedded system A lot of merits by the software sequential control system can be expected in the respect of economic reproduction, intelligent technologies and utilities, And porting of LINUX operating system to embedded system is carried out and device drivers and interface boards for LINUX 05 are designed for controlling air compressor by software Internet remote control and monitoring system of air compressor is implemented with Java script and CGI for these purposes. The experiment for operating air compressor system is taken through internet networks. The results show that developed system can be used for real plant.

• 제어로봇시스템학회논문지
• /
• 제10권2호
• /
• pp.171-177
• /
• 2004

Many instrumentations have been used to acquire the performance data of military systems fer many years. But they could not satisfy environmental specifications(vibration, shock, temperature) and processing speed to apply for the performance test of military systems because of having developed as common vehicles/fixed installation equipments. Thus a new rugged embedded measurement system is required to process large data in high processing speed(Maximum sample rate:1.25Mhz/ch) with rugged environmental specifications. We have developed embedded measurement systems by using PXI(PCI extension for Instrumentation)bus interface composed of a stand alone controller and versatile data acquisition boards(analog, digital, vision, temperature and small signal conditioner) on PC-based environment to solve these problems. Operation programs have been developed using Lab_View and the performances have been validated experimentally.

• 대한전기학회논문지:시스템및제어부문D
• /
• 제54권3호
• /
• pp.141-147
• /
• 2005

In the field of electrical industry, embedded computing environment including hardware and software is getting more important as the industry shifts to the knowledge-based one. Java could play a great role as bridging technology in such a transition because it provides a lot of benefits like dynamic application download, compatibility of cross platform, and its own security solution. However, the Java technology has a limitation of real-time problem when it is applied to the embedded computing system of the electrical industry. To solve the problem, a novel java-native combination model has been proposed and designed to a firmware level. This scheme has been employed in four kinds of control boards. The result shows that the proposed model has great potential to implement the real-time processing in control of the devices.

• 로봇학회논문지
• /
• 제17권2호
• /
• pp.133-141
• /
• 2022

This paper introduces model compression algorithms which make a deep neural network smaller and faster for embedded systems. The model compression algorithms can be largely categorized into pruning, quantization and knowledge distillation. In this study, gradual pruning, quantization aware training, and knowledge distillation which learns the activation boundary in the hidden layer of the teacher neural network are integrated. As a large deep neural network is compressed and accelerated by these algorithms, embedded computing boards can run the deep neural network much faster with less memory usage while preserving the reasonable accuracy. To evaluate the performance of the compressed neural networks, we evaluate the size, latency and accuracy of the deep neural network, DenseNet201, for image classification with CIFAR-10 dataset on the NVIDIA Jetson Xavier.