• Proceedings of the Korean Society of Propulsion Engineers Conference
• /
• 2012.05a
• /
• pp.167-171
• /
• 2012

A computational analysis of nozzle flow is conducted to investigate effects of the flight altitude on thrust performance. Reynolds-averaged Navier-Stokes equation with k-${\omega}$ SST(Shear Stress Transport) turbulence model is employed to simulate the nozzle flow in various altitude conditions, where continuum mechanics is to be valid. Thrust performance of the nozzle is exceedingly poor upto 10 km of flight altitude because of the irreversible phenomena such as shock and/or flow separation occurring inside the nozzle, whereas it is restored to the nominal value as the altitude is attained higher than 30 km.

• Proceedings of the KSR Conference
• /
• 2008.11b
• /
• pp.310-315
• /
• 2008

Authors design a rotary-type small-scaled linear induction motor(LIM) and a performance test machine before manufacturing a real-scaled LIM for a railway transit. The designed LIM is a single-sided, short primary type and its primary has 4 poles. The rated power is 10(kW). In order to analyze characteristics of the LIM, authors use a mixed 2D-3D FEM analysis. 3-D FEM analysis is used for calculating a distribution of eddy-current on the semi-caped AL-sheet of the secondary reaction. Authors calculate a correction factor of conductivity and an equivalent conductivity on the secondary AL-sheet using a normalized eddy-current. The equivalent conductivity which is calculated in this way includes a transverse edge-effect of the LIM. Authors apply the equivalent conductivity on the secondary AL-sheet of 2D-FEM model and get performance characteristics of the LIM. Basic characteristics such as thrust and normal force, input current, efficiency and power factor of the LIM have been analyzed with the variation of frequency and speed. In order to apply an air-gap control system, the variation of the basic characteristics have been analyzed with the air-gap length variation of the LIM. Finally, authors introduce an operation mode using the air-gap control system and conduct a research on feasibility of the system.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2021.06a
• /
• pp.368-368
• /
• 2021

기후 변화로 인한 극한사상의 크기와 빈도 변화를 예측하는 것은 수공 인프라 설계에 있어 주된 관심사 중 하나이다. 보통 극한사상에 대한 강도, 빈도, 지속시간에 대한 정보가 필요하며, 이는 일반적으로 IDF(Intensity-Duration-Frequency) 곡선으로부터 추출된다. 최근 CMIP(Coupled Model Intercomparison Project) 6단계에서 새로운 이산화탄소 배출 시나리오와 업데이트된 기후모델을 이용하여 미래의 기후에 대한 예측 시계열을 발표했으므로, 미래 기후 변화 시나리오를 기반으로 IDF 곡선을 새로 추정하고 미래 기간의 변화를 평가할 필요가 있다. 본 연구에서는 한국의 40개 지역에 대해 일단위 자료를 시단위로 축소(downscaling)한 후, 확률론적 일기생성기(stochastic weather generator)를 이용하여 30년 시단위 시계열을 100개의 앙상블로 생성하였다. 생성된 시계열로부터 연최대강수량 시계열을 재구성하여 GEV 분포와 gumbel 분포에 적용하였다. 적합도 검정(Anderson-Darling(AD) 검정 및 Kolmogorov-Smirnov(KS) 검정)을 수행하였으며, 과거 자료를 기반으로 생성된 IDF 곡선과 비교 검증하였다. CMIP5의 기후변화 자료를 사용한 결과와 CMIP6 기후변화의 결과를 비교하였으며, 본 연구의 주요 결과는 다음과 같다. (1) 향후 강우 강도는 증가할 것이며 강우 강도의 증가는 말기에 현저하게 관찰될 것이다. (2) 시간별 강우 강도의 미래 변화가 일단위 강우 강도보다 더 크다. (3) 강우 강도의 불확실성을 정량화하기 위해 앙상블을 사용해야 한다. (4) 강우 강도의 미래 변화에 대한 공간적인 경향이 확인된다. 시단위 시계열 앙상블을 생성하여 추정된 IDF 곡선에 대한 정보는 기후 변화의 영향을 평가하고 적절한 적응 및 대응 전략을 개발하는 데 도움이 될 것이다.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2020.07a
• /
• pp.664-666
• /
• 2020

데이터의 다양성은 학습에 따른 모델의 성능을 좌지우지하는 중요한 요소이다. 그렇기 때문에 많은 양의 데이터를 확보하는 것은 학습에 있어서 아주 중요하다. 하지만, 데이터를 수집하는 것은 시간과 비용이 많이 드는 단계 중 하나이다. 본 논문에서는 제한된 데이터를 가지고 이미지 처리를 거쳐 대량의 데이터로 증폭시켜 많은 양의 데이터를 확보하는 과정에 대해 제안한다. 가지고 있는 YOLOv4용 학습 데이터 셋을 활용하여 사용자로부터 입력받은 확대/축소 비율, 각도로 데이터를 변형하고, 이렇게 추가로 생성된 데이터 셋을 기존 학습 데이터 셋에 재포함시키는 소프트웨어를 개발하는 것을 목표로 한다. 구현된 소프트웨어로 증폭된 대량의 데이터 셋을 다시 원본 학습 데이터 셋에 추가하고, 같은 영상에 대해서 원본 데이터 셋만 학습시킨 경우의 객체 검출 결과와 증폭된 학습 데이터 셋이 포함된 데이터 셋의 경우의 객체 검출 결과를 비교하여 그 성능을 검증하고 분석하도록 한다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2008.11a
• /
• pp.1563-1566
• /
• 2008

본 논문에서는 데이터의 전처리과정으로 SNMP MIB 데이터에 대한 속성 부분집합의 선택 방법(attribute subset selection)을 사용하여 특징선택 및 축소(feature selection & reduction)를 실시하였다. 또한 데이터 마이닝의 대표적인 해석학적 분석 모델인 연관관계규칙기법(association rule mining)을 이용하여 트래픽 폭주 공격 및 공격유형별 SNMP MIB 데이터에 내재되어 있는 특징들을 규칙의 형태로 추출하여 분석하는 의미론적 심층해석을 실시하였다. 공격유형에 대한 패턴 규칙의 추출 및 분석은 공격이 발생한 프로토콜에 대해서만 서비스를 제한하고 관리할 수 있는 정책적 근거를 제공함으로써 보다 안정적인 네트워크 환경과 원활한 자원관리를 지원할 수 있다. 본 논문에서 제시한 트래픽 폭주 공격 및 공격유형별 데이터로부터의 자동적 특징의 규칙 추출 및 의미론적 해석방법은 침입탐지 시스템을 위한 새로운 방법론에 모멘텀을 제시할 수 있다는 긍정적인 가능성과 함께 침입탐지 및 대응시스템의 정책 수립을 지원할 수 있을 것으로 기대된다.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.21 no.7
• /
• pp.285-291
• /
• 2020

In manufacturing, humans are being replaced with robots, but expert skills remain difficult to convert to data, making them difficult to apply to industrial robots. One method is by visual motion recognition, but physical features may be judged differently depending on the image data. This study aimed to improve the accuracy of vision methods for estimating the posture of humans. Three OpenPose vision models were applied: MPII, COCO, and COCO+foot. To identify the effects of face-covering accessories and image preprocessing on the Convolutional Neural Network (CNN) structure, the presence/non-presence of accessories, image size, and filtering were set as the parameters affecting the identification of a human's posture. For each parameter, image data were applied to the three models, and the errors between the actual and predicted values, as well as the percentage correct keypoints (PCK), were calculated. The COCO+foot model showed the lowest sensitivity to all three parameters. A <50% (from 3024×4032 to 1512×2016 pixels) reduction in image size was considered acceptable. Emboss filtering, in combination with MPII, provided the best results (reduced error of <60 pixels).

• Transactions of the Korean Society of Mechanical Engineers B
• /
• v.36 no.2
• /
• pp.189-196
• /
• 2012

The heavy moderator acts as the ultimate heat-sink in an operating CANDU reactor. HUKINS has been developed to investigate moderator flow patterns. HUKINS consists of a 38.4-mm-thick cylindrical shell with a 0.95 m inner diameter and 88 sus-tubes that produce a total heat of 10 kW. A chemical visualization method was selected to estimate the occurrence of typical moderator flow patterns. Momentum-dominated flow, mixed flow, and buoyancy-dominated flow are detected under conditions of a heat load of 7.7 kW and input mass flow rates of 4, 7, and 11 L/min. The experimental results are similar to the results of a CFD simulation that consisted of approximately 1.9 million grids and was conducted using the k-${\varepsilon}$ turbulence model. Therefore, both the present experiments and simulations using HUKINS, a 1/8-scale model, represent all three important flow patterns expected in the real CANDU6 reference reactor. Thus, it has been demonstrated that HUKINS could be useful in the study of CANDU6 moderator circulation.

• Journal of the Korea Society of Computer and Information
• /
• v.27 no.2
• /
• pp.33-42
• /
• 2022

In this paper, we proposed a deep learning model based on multi-scale dilated convolution feature fusion for the segmentation of black ice region in road image to send black ice warning to drivers in real time. In the proposed multi-scale dilated convolution feature fusion network, different dilated ratio convolutions are connected in parallel in the encoder blocks, and different dilated ratios are used in different resolution feature maps, and multi-layer feature information are fused together. The multi-scale dilated convolution feature fusion improves the performance by diversifying and expending the receptive field of the network and by preserving detailed space information and enhancing the effectiveness of diated convolutions. The performance of the proposed network model was gradually improved with the increase of the number of dilated convolution branch. The mIoU value of the proposed method is 96.46%, which was higher than the existing networks such as U-Net, FCN, PSPNet, ENet, LinkNet. The parameter was 1,858K, which was 6 times smaller than the existing LinkNet model. From the experimental results of Jetson Nano, the FPS of the proposed method was 3.63, which can realize segmentation of black ice field in real time.

• Journal of the Korean Geosynthetics Society
• /
• v.22 no.2
• /
• pp.1-12
• /
• 2023

Research on the causes of landslides and prediction of vulnerable areas is being conducted globally. This study aims to predict the effective soil depth, a critical element in analyzing and forecasting landslide disasters, using topographic information. Topographic data from various institutions were collected and assigned as attribute information to a 100 m × 100 m grid, which was then reduced through data grading. The study predicted effective soil depth for two cases: three depths (shallow, normal, deep) and five depths (very shallow, shallow, normal, deep, very deep). Three classification models, including K-Nearest Neighbor, Random Forest, and Deep Artificial Neural Network, were used, and their performance was evaluated by calculating accuracy, precision, recall, and F1-score. Results showed that the performance was in the high 50% to early 70% range, with the accuracy of the three classification criteria being about 5% higher than the five criteria. Although the grading criteria and classification model's performance presented in this study are still insufficient, the application of the classification model is possible in predicting the effective soil depth. This study suggests the possibility of predicting more reliable values than the current effective soil depth, which assumes a large area uniformly.

• Journal of the Computational Structural Engineering Institute of Korea
• /
• v.26 no.2
• /
• pp.165-171
• /
• 2013

Due to the difficulty in considering dynamic load in the view point of a computer resource and computing time, it is common that external load is assumed as ideal static loads. However, structural analysis under static load cannot guarantee the safety of design of the structures under dynamic loadings. Recently, the systematic method to construct equivalent static load from the given dynamic load has been proposed. Previous study has calculated equivalent static load through the optimization procedure under displacement constraints. However, previously reported works to distribute equivalent static load were based on ad-hoc methods. Improper selection of equivalent static loading positions may results in unreliable prediction of structural design. The present study proposes the selection method of the proper locations of equivalent static loads to dynamically applied loads when we consider transient dynamic structural problems. Moreover, it is appropriate to take into account the stress constraint as well as displacement constraint condition for the safety design. But the previously reported studies of equivalent static load design methods considered only displacement constraint conditions but not stress constraint conditions. In the present study we consider not only displacement constraint but also stress constraint conditions. Through a few numerical examples, the efficiency and reliability of proposed scheme is verified by comparison of the equivalent stress between equivalent static loading and dynamic loading.