• Nuclear Engineering and Technology
• /
• 제52권7호
• /
• pp.1429-1435
• /
• 2020

Applying an accurate parametric prediction model to identify abnormal or false pressurizer water levels (PWLs) is critical to the safe operation of marine pressurized water reactors (PWRs). Recently, deep-learning-based models have proved to be a powerful feature extractor to perform high-accuracy prediction. However, the effectiveness of models still suffers from two issues in PWL prediction: the correlations shifting over time between PWL and other feature parameters, and the example imbalance between fluctuation examples (minority) and stable examples (majority). To address these problems, we propose a cost-sensitive mechanism to facilitate the model to learn the feature representation of later examples and fluctuation examples. By weighting the standard mean square error loss with a cost-sensitive factor, we develop a Cost-Sensitive Long Short-Term Memory (CSLSTM) model to predict the PWL of PWRs. The overall performance of the CSLSTM is assessed by a variety of evaluation metrics with the experimental data collected from a marine PWR simulator. The comparisons with the Long Short-Term Memory (LSTM) model and the Support Vector Regression (SVR) model demonstrate the effectiveness of the CSLSTM.

• 한국환경복원기술학회지
• /
• 제15권6호
• /
• pp.17-27
• /
• 2012

Climate is generally accepted as one of the major determinants of plants distribution. Plants are sensitive to bioclimates, and local variations of climate determine habitats of plants. The purpose of this paper is to identify the factors affecting the distribution of narrow-range plants in South Korea using National Survey of Natural Environment data. We developed species distribution models for 6 plant species using climate, topographic and soil factors. All 6 plants were most sensitive to climatic factors but less other factors at national scale. Meliosma myriantha, Stewartia koreana and Eurya japonica, distributed at southern and coast region in Korea, were most sensitive to precipitation and temperature. Meliosma myriantha was mostly effected by annual precipitation and precipitation of driest quarter, Stewartia koreana was effected by annual precipitation and elevation, and Eurya japonica was affected by temperature seasonality and precipitation of driest quarter. On the other hand, Spiraea salicifolia, Rhododendron micranthum and Acer tegmentosum, distributed at central and northern inland in Korea, were most sensitive to temperature and elevation. Spiraea salicifolia was affected by mean temperature of coldest quarter and annual mean temperature, Rhododendron micranthum and Acer tegmentosum were affected by mean temperature of warmest quarter and elevation. We can apply this result to future plant habitat distribution under climate change.

• 한국컴퓨터정보학회논문지
• /
• 제29권5호
• /
• pp.93-100
• /
• 2024

4차 산업혁명의 도래와 함께 인공지능, 빅데이터, 사물인터넷, 클라우드 컴퓨팅 등의 첨단 정보 기술이 다양한 산업 분야에서 혁신을 이끌고 있다. 이 기술들은 방대한 양의 데이터를 생성하고 있으며, 많은 기업들이 이를 활용하고 있다. 그러나 개인 데이터 수집 시 발생할 수 있는 프라이버시 침해 위험으로 인해 사용자들은 민감한 정보 제공을 망설이고 있다. 특히 의료 분야에서는 환자의 병명과 같은 민감한 정보 수집이 큰 도전이 되고 있으며, 프라이버시 문제가 데이터 수집과 분석의 장애가 되고 있다. 본 연구는 프라이버시 보호를 유지하면서도 통계적 정보를 효과적으로 추출할 수 있는 의료 데이터 수집 및 분석 기법을 제안한다. 제안 기법은 기존의 단순한 데이터 수집을 넘어서, 프라이버시를 보장하면서 수집된 데이터에서 통계적 정보를 효과적으로 추출하는 방법을 포함한다. 실제 데이터를 이용한 성능 평가에서는 제안된 기법이 기존 방법보다 더 효과적으로 프라이버시를 보존하며 통계적 정보를 도출할 수 있음을 입증한다.

• 한국농공학회논문집
• /
• 제60권3호
• /
• pp.15-25
• /
• 2018

Estimating the reference evapotranspiration is an important factor to consider in irrigation system design and agricultural water use. However, there is a limitation in using the FAO Penman-Monteith (FAO P-M) equation, which requires various meteorological data. The purpose of this study is to compare three reference evapotranspiration (ETo) equations in the case of meteorological data missing for 11 study weather stations. Firstly, the FAO P-M equation is used for reference potential evapotranspiration estimation with the actual solar radiation data $R_n$ and the actual vapor pressure $e_a$. Then, in the case of $R_n$, and $e_a$ are missed, the reference evapotranspirations applying FAO P-M, Priestley-Taylor (P-T), Hargreaves (HG) equation were calculated using other meteorological factors. Secondly, MAE, RMSE, $R^2$ were calculated to compare ETo relationship from the ETo equations. From the results, ETo with Hargreaves equation in coastal areas and the Priestley-Taylor equation in the inland areas showed relatively high correlation with FAO P-M when $e_a$ data is missed. In the case of $R_n$ data is missed or two weather data, $e_a$, and $R_n$ data are all missed, $R^2$ value in Priestley-Taylor equation was highest in coastal areas, and $R^2$ values in Hargreaves equation were the high values for 7 inland areas. The results of sensitivity analysis showed that net radiation was the most sensitive for P-T and HG equation, and for FAO P-M, the most sensitive factor was net radiation and relative humidity, air temperature and wind speed were follows. Therefore, in considering of the accessibility to the coast, the types of the missing wether data, and the correlation and the magnitude of error, the reference evapotranspiration equations would be selected in sense of different conditions.

• 정보과학회 논문지
• /
• 제42권7호
• /
• pp.919-928
• /
• 2015

최근 대용량 데이터 대상의 프라이버시 보호 데이터 마이닝(privacy-preserving data mining: PPDM)이 활발히 연구되고 있다. 본 논문에서는 민감한 데이터 집합의 상관관계를 파악하는데 널리 사용되는 주성분 분석 기반의 PPDM을 다룬다. 일반적으로 주성분 분석은 모든 데이터를 한 곳에 모아 처리해야 하므로 민감한 데이터가 서로에게 공개되고, 상당한 계산량을 요구하며, 또한 데이터를 모으는 과정에서 많은 통신 오버헤드가 발생한다. 이러한 문제를 해결하기 위하여 본 논문은 데이터를 한 곳에 모으지 않고도 주성분 분석을 안전하게 계산하는 효율적인 방법을 제안한다. 제안하는 방법은 노드들 간에 한정된 정보만을 공유하면서도 원래의 주성분 분석 결과와 동일한 결과를 얻을 수 있다. 또한 안전한 주성분 분석에 저차원 변환을 적용하여 안전한 유사 문서 검색에 사용한다. 마지막으로 다양한 실험을 통해 제안한 방법이 대용량의 다차원 데이터에서 효율적으로 동작함을 확인한다.

• International Journal of Aeronautical and Space Sciences
• /
• 제11권3호
• /
• pp.193-200
• /
• 2010

Through experimental investigations utilizing hypersonic shock tunnel-coaxial thermocouples as well as blow down hypersonic wind tunnel-temperature sensitive paints, the heat flux and the temperature over a protuberance were measured and analyzed. The experimental data were subsequently compared to heat flux data that was obtained by using blow down hypersonic wind tunnel and heat flux gauges. According to the comparison, both sets of data illustrated correlation with one another. The measured heat flux was large when the height of the protuberance was large. Experimental results show that heat flux measurements taken at higher locations were greater than those taken at lower locations. For high protuberances, a severe jump in the heat flux was observed, ranging in values within 0.6-0.7 of the height of the protuberances. However, when the protuberance was sufficiently short, a rise in the heat flux was rarely observed as the protuberance was totally submerged under the separation region.

• 한국환경과학회지
• /
• 제25권7호
• /
• pp.951-961
• /
• 2016

To prevent increasing instances of heat-related illnesses due to heat waves generated by climate change, a customized thermal environment index should be developed for outdoor workers. In this study, we conducted sensitivity analysis of the Masan harbor during a heat wave period (August 9th to 15th, 2013) using the MENEX model with metabolic rate and clothing-insulation data, in order to obtain realistic information about the thermal environment. This study shows that accurate input data are essential to gather information for thermophysiological indices (PST, DhR, and OhR). PST is sensitive to clothing insulation as a function of clothing. OhR is more sensitive to clothing insulation during the day and to the metabolic rate at night. From these results, it appears that when exposed to high-temperature thermal environments in summer, wearing highly insulated clothing and getting enough rest (to lower the metabolic rate) can aid in preventing heat-related illnesses. Moreover, in the case of high-intensity harbor work, quantification of allowed working time (OhR) during heat waves is significant for human health sciences.

• 대한교통학회지
• /
• 제7권1호
• /
• pp.35-42
• /
• 1989

The multinomial logit model has been applied for various choice problems. Among others, the joint destination mode choice, the mode choice and the route choice are the three major modeling topics for korean transportation planners. This paper examines with real world data (the Olympic road and its competing two major arterials) the usefulness of a Logit route choice model. Quites surpisingly, it is found that the multinomial route choice behavioral model calibrated for this study based on (0,1) individula data base can not provide a good estimate for O-D trips less than 6㎞. 400data points and 3case studies might not be sufficient for a sound conclusion. It is, however, believed from a series of similar studies conducted by the authors that the route choice behavior is more sensitive (more demand elastic with respect to travel time changes) than the mode choice and the shorter trip, the more sensitive. The travel time parameters for destination choice models are usually smalle than the travel time parameters for mode choice models and these parameters (for mode choice models) turn our smaller than the travel time parameters for route choice models from this study. Table 2 in this paper shows parameter changes for three different markets and Table 3 shows the modeling errors when the estimated individual probabilities are aggregated into a route level.

• 한국지능시스템학회논문지
• /
• 제15권2호
• /
• pp.204-210
• /
• 2005

본 논문에서는 TSK 퍼지 모델을 위한 온라인 식별 알고리즘을 제안한다. 제안된 알고리즘은 거리를 이용하여 TSK 퍼지 모델에 대한 전건부의 구조를 식별하고, 재귀적 최소자승법으로 후건부를 구성하는 부분 선형 함수들의 매개 변수를 구한다. 대부분의 다른 연구들에서는 전건부의 구조를 구하기 위해서 클러스터링을 수행할 때 입력 공간에서만 고려하였으나. 제안된 알고리즘에서는 입력 공간 및 출력 공간 모두에서 고려하여, 아웃라이어를 효과적으로 배제할 수 있다. 기존의 대부분의 다른 알고리즘에서 샘플 데이터자체를 클러스터의 중심으로 사용하여 잡음에 민감한 단점이 있었으나, 제안된 알고리즘에서는 데이터 자체를 클러스터의 중심으로 사용하지 않아 잡음에 대해 민감하지 않다. 제안된 알고리즘은 많은 데이터의 저장을 필요로 하지 않고, 한 번 통과함으로써 모델을 구할 수 있다.

• 한국통신학회논문지
• /
• 제34권11B호
• /
• pp.1169-1177
• /
• 2009

본 논문은 센서 네트워크상에서 지연시간에 민감한 데이터 전송을 위한 에너지 효율적인 MAC 프로토콜을 제안한다. 비 동기식 센서 네트워크에서는 소비되는 에너지와 지연시간이 채널 모니터링 주기(Channel Monitoring Interval)와 데이터 센싱 주기(Sensing Data Interval)에 따라 변화한다. 이러한 사실을 기반으로 본 논문에서는 오버히어링(Overhearing)으로 발생하는 에너지 소비를 줄이기 위해 자신의 채널 모니터링 주기를 이웃노드에 알리는 새로운 프리엠블 구조와 낮은 연산 복잡도를 가지는 채널 모니터링 주기를 결정하는 알고리즘을 제안한다. 실험결과에서는 다른 센서 MAC 프로토콜과 제안한 MAC 프로토콜과의 성능비교를 보여준다.