• Proceedings of the Korean Information Science Society Conference
• /
• 2001.04b
• /
• pp.586-588
• /
• 2001

최근에는 순환 신경망의 생성모델이 비교사 학습에 관련하여 활발히 연구되고 있다. 이러한 형태의 신경망은 형태 추출이나 인식에 효과적으로 사용될 수 있는 반면 반복 loop를 사용하므로 대단히 많은 계산이 필요하다. 본 논문에서는 Oh와 Seung에 의해 제안된 상향전파(Up-propagation) network이라는 순환 신경망을 FPGA를 이용해서 구현하였다. 단층 신경망은 9개의 상층 neuron과 256개의 하층 neuron으로 구성되 있으며 4만 게이트의 FPGA 하나로 효과적으로 구현할 수 있다. pipeline된 곱셈기로 게산 속도를 향상시켰고 sigmoid 전달 함수는 유한 정밀도의 2차 다항식으로 근사될 수 있다. 구현된 하드웨어는 hand-written 숫자 영상인 USPS data를 재생하는데 사용되었으며 좋은 결과를 얻었다.

• 한국가스학회:학술대회논문집
• /
• 2003.10a
• /
• pp.107-113
• /
• 2003

• 한국가스학회:학술대회논문집
• /
• 2005.10a
• /
• pp.5-11
• /
• 2005

• Journal of the Korean Society for Nondestructive Testing
• /
• v.35 no.1
• /
• pp.31-38
• /
• 2015

Intelligent leak detection is an essential component of a underground water supply pipeline network such as a smart water grid system. In this network, numerous leak detection sensors are needed to cover all of the pipelines in a specific area installed at specific regular distances. It is also necessary to determine the existence of any leaks and estimate its location within a short time after it occurs. In this study, the leak signal properties and feasibility of leak location detection were investigated when concurrent leaks occurred at two points in a pipeline. The straight distance between the two leak sensors in the 100A sized cast-iron pipeline was 315.6 m, and their signals were measured with one leak and two concurrent leaks. Each leak location was described after analyzing the frequency properties and cross-correlation of the measured signals.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.14 no.3
• /
• pp.1086-1103
• /
• 2020

The automatic identification and classification of image-based weld defects is a difficult task due to the complex texture of the X-ray images of the weld defect. Several depth learning methods for automatically identifying welds were proposed and tested. In this work, four different depth convolutional neural networks were evaluated and compared on the 1631 image set. The concavity, undercut, bar defects, circular defects, unfused defects and incomplete penetration in the weld image 6 different types of defects are classified. Another contribution of this paper is to train a CNN model "RayNet" for the dataset from scratch. In the experiment part, the parameters of convolution operation are compared and analyzed, in which the experimental part performs a comparative analysis of various parameters in the convolution operation, compares the size of the input image, gives the classification results for each defect, and finally shows the partial feature map during feature extraction with the classification accuracy reaching 96.5%, which is 6.6% higher than the classification accuracy of other existing fine-tuned models, and even improves the classification accuracy compared with the traditional image processing methods, and also proves that the model trained from scratch also has a good performance on small-scale data sets. Our proposed method can assist the evaluators in classifying pipeline welding defects.

• Journal of Korean Society on Water Environment
• /
• v.23 no.1
• /
• pp.103-110
• /
• 2007

An efficient operational strategy using expert system for metro water supply systems in case of emergency situations is developed in this study. The emergency situations of the water supply systems are classified into three categories : pipeline system accident, machinery and electric facility accident and water quality accident. A PC-based expert system is developed using CLIPS for Seoul metro water supply system, Phase 1 & 2 system and Phase 3 & 4 system. Broad professional knowledges and experiences from the experts in the water supply systems have been collected systematically to construct the knowledge base. Decision-making in case of an emergency is based upon the professional knowledge so that a rational and efficient operational management can be available even in the absence of experienced expert. Especially the expert model developed in this study also provides a guide for pumping operation in case of pipeline accident to confirm that the proper pressure to all nodes in the system is supplied. The pipe network simulator KYPIPE has been consecutively executed by trial and error fashion for each pipeline in the system. The results from KYPIPE were included in the knowledge base to supplement the knowledge of the field engineers.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.11 no.5
• /
• pp.2433-2452
• /
• 2017

Recently, the pipeline-forwarding has been proposed as a new technique to resolve the end-to-end latency problem of the duty-cycle MAC protocols in Wireless Sensor Networks (WSNs). Some protocols based on this technique such as PMAC and PRI-MAC have shown an improvement not only in terms of reducing end-to-end latency but also in terms of reducing power consumption. In these protocols, however, the sensor nodes still waste a significant amount of energy for unnecessary idle listening during contention period of upstream nodes to check the channel activity. This paper proposes a new pipeline-forwarding duty-cycle MAC protocol, named RP-MAC (Reduced Pipelined duty-cycle MAC), which tries to reduce the waste of energy. By taking advantage of ACK mechanism and shortening the handshaking procedure, RP-MAC minimizes the time for checking the channel and therefore reduces the energy consumption due to unnecessary idle listening. When comparing RP-MAC with the existing solution PRI-MAC and RMAC, our QualNet-based simulation results show a significant improvement in term of energy consumption.

• Journal of the Korean Society of Safety
• /
• v.39 no.1
• /
• pp.27-32
• /
• 2024

This study analyzed cases of hydrogen (H₂) and natural gas (CH₄) leakage from a hydrogen-blended natural gas pipeline to determine a range of leakage characteristics, including leakage type, pipe material, pipe diameter, pressure, and damage size. Based on the results of this analysis, five hydrogen-blended natural gas leakage scenarios were selected. The national vision for a carbon-neutral society by 2050 is a very important strategic objective and promotes environmentally sustainable economic development in the age of the climate crisis. Accordingly, zero-carbon and low-carbon policies are being promoted in various fields, including energy production, consumption, and industrial processes. Hydrogen-blended natural gas is eco-friendly and is considered an important step towards carbon neutrality, with various countries including the United States and several European countries conducting empirical research to further investigate its potential. In Korea, a national research project commenced in April 2023 to verify and demonstrate the life cycle safety of blending hydrogen into the natural gas network. The results of this study will provide important data for the analysis of the damage impacts caused by the leakage of hydrogen-blended natural gas, such as the diffusion of gas clouds, fires, and gas explosions.

• Journal of Energy Engineering
• /
• v.17 no.3
• /
• pp.139-144
• /
• 2008

Recently due to march as the high oil price, It is necessary for Korea to grope a plan, which is to increase the energy efficiency of existing facilities as well as to develop overseas gas and oil resources. With this point, this work carried out to approach the high pressure LNG pipeline network of Inchon receiving terminal with Newton method as corrective flowrate. We found that the high pressure network mainly depends on FCVs(Flow Control Valves). The high pressure pump showed the maximum efficiency at the FCVs of 50% opening and could discharge LNG only above the LNG head of 1,500m from a system curve obtained. The operating cost of pumps was estimated from their operating points. We compared the operating cost under normal operation with the operating cost under maximum efficiency. Especially, we obtained the day savings of a year as wells as the hour savings of a day. From the results, the high pressure network win be able to reduce the operating cost of 138 million wons in a year. This means that a pump can reduce the operating cost of 9,823 thousands won. Consequently, this work could find the operating features of the pumps under the complicated high pressure LNG network and the savings effect of the pump operating cost. Also, the results will be able to macroscopically contribute the heightening of national energy competitiveness as well as to microscopically contribute the future effective operation of LNG receiving terminal.

• Journal of Korean Society of Water and Wastewater
• /
• v.34 no.2
• /
• pp.139-147
• /
• 2020

This paper suggests a nonlinear pressure consideration scheme through an unsteady pipe network analyzer for leakage detection with a portable pressure wave generator. In order to evaluate the performance of a proposal scheme, linear input pattern has been simulated and experiments had been carried out under both no leakage and one leakage conditions in a reservoir-pipeline-valve system. This method using portable pressure wave generator showed that a leakage can be detected from a reflection where a leakage is originated through time domain analysis. Meaningful similarity in pressure response between nonlinear input pattern and experimental results were found both no leakage and a leakage conditions.