• 시스템엔지니어링학술지
• /
• 제18권2호
• /
• pp.94-107
• /
• 2022

Accidents prevention and mitigation is the highest priority of nuclear power plant (NPP) operation, particularly in the aftermath of the Fukushima Daiichi accident, which has reignited public anxieties and skepticism regarding nuclear energy usage. To deal with accident scenarios more effectively, operators must have ample and precise information about key safety parameters as well as their future trajectories. This work investigates the potential of machine learning in forecasting NPP response in real-time to provide an additional validation method and help reduce human error, especially in accident situations where operators are under a lot of stress. First, a base-case SGTR simulation is carried out by the best-estimate code RELAP5/MOD3.4 to confirm the validity of the model against results reported in the APR1400 Design Control Document (DCD). Then, uncertainty quantification is performed by coupling RELAP5/MOD3.4 and the statistical tool DAKOTA to generate a large enough dataset for the construction and training of neural-based machine learning (ML) models, namely LSTM, GRU, and hybrid CNN-LSTM. Finally, the accuracy and reliability of these models in forecasting system response are tested by their performance on fresh data. To facilitate and oversee the process of developing the ML models, a Systems Engineering (SE) methodology is used to ensure that the work is consistently in line with the originating mission statement and that the findings obtained at each subsequent phase are valid.

• 한국환경복원기술학회지
• /
• 제25권3호
• /
• pp.29-45
• /
• 2022

For biodiversity conservation, the importance of beta-diversity which is changes in the composition of species according to environmental changes has become emphasized. However, given the systematic investigation of species distribution and the accumulation of large amounts of data in the Republic of Korea(ROK), research on the spatialization of beta-diversity using them is insufficient. Accordingly, this research investigated the applicability of the Generalized Dissimilarity Modeling(GDM) to ROK, which can predict and map the similarity of compositional turnover (beta-diversity) based on environmental variables. A brief overview of the statistical description on using GDM was presented, and a model was fitted using the flora distribution data(410,621points) from the National Ecosystem Survey and various environmental spatial data including climate, soil, topography, and land cover. Procedures and appropriated spatial units required to improve the explanatory power of the model were presented. As a result, it was found that geographical distance, temperature annual range, summer temperature, winter precipitation, and soil factors affect the dissimilarity of the vegetation community composition. In addition, as a result of predicting the similarity of vegetation composition across the nation, and classifying them into 20 and 100 zones, the similarity was high mainly in the central inland area, and tends to decrease toward the mountainous areas, southern coastal regions, and island including Jeju island, which means the composition of the vegetation community is unique and beta diversity is high. In addition, it was identified that the number of common species between zones decreased as the geographic distance between zones increased. It classified the spatial distribution of plant community composition in a quantitative and objective way, but additional research and verification are needed for practical application. It is expected that research on community-level biodiversity modeling in the ROK will be conducted more actively based on this study.

• 운동영양학회지
• /
• 제25권1호
• /
• pp.35-41
• /
• 2021

[Purpose] This pilot study aimed to develop a regression model to estimate the excess post-exercise oxygen consumption (EPOC) of Korean adults using various easy-to-measure dependent variables. [Methods] The EPOC and dependent variables for its estimation (e.g., sex, age, height, weight, body mass index, fat-free mass [FFM], fat mass, % body fat, and heart rate_sum [HR_sum]) were measured in 75 healthy adults (31 males, 44 females). Statistical analysis was performed to develop an EPOC estimation regression model using the stepwise regression method. [Results] We confirmed that FFM and HR_sum were important variables in the EPOC regression models of various exercise types. The explanatory power and standard errors of estimates (SEE) for EPOC of each exercise type were as follows: the continuous exercise (CEx) regression model was 86.3% (R²) and 85.9% (adjusted R²), and the mean SEE was 11.73 kcal, interval exercise (IEx) regression model was 83.1% (R²) and 82.6% (adjusted R²), while the mean SEE was 13.68 kcal, and the accumulation of short-duration exercise (AEx) regression models was 91.3% (R²) and 91.0% (adjusted R²), while the mean SEE was 27.71 kcal. There was no significant difference between the measured EPOC using a metabolic gas analyzer and the predicted EPOC for each exercise type. [Conclusion] This pilot study developed a regression model to estimate EPOC in healthy Korean adults. The regression model was as follows: CEx = -37.128 + 1.003 × (FFM) + 0.016 × (HR_sum), IEx = -49.265 + 1.442 × (FFM) + 0.013 × (HR_sum), and AEx = -100.942 + 2.209 × (FFM) + 0.020 × (HR_sum).

• 문화기술의 융합
• /
• 제7권4호
• /
• pp.27-34
• /
• 2021

본 연구는 중년여성의 성공적 노화와 SOC 전략이 삶의 질에 미치는 영향을 파악하기 위한 서술적 조사연구이다. 연구대상자는 전국의 40~65세의 중년여성으로 구조화된 설문지를 이용하여 자료를 수집하였다. 수집된 자료는 SPSS WIN/PC 24.0통계프로그램을 이용해 서술적 통계, Independent t-test, one-way ANOVA, Pearson's correlation, stepwise multiple regression으로 분석하였다. 연구결과 중년여성의 성공적인 노화, SOC 대체전략과 삶의 질은 정적인 상관관계가 있었으며, 삶의 질에 미치는 요인으로 성공적 노화(β=.41, p<.001), SCO 대처전략(β=.17, p=.047) 순으로 통계적 유의성이 확인되었다. 두 변수에 의한 설명력은 24.7%였다. 따라서 중년여성의 삶의 질을 향상시키기 위해서는 성공적 노화와 SCO 대처전략을 증진시키는 프로그램의 개발과 적용이 필요하다.

• Wind and Structures
• /
• 제36권2호
• /
• pp.121-131
• /
• 2023

Aiming at the problem that fatigue characteristics of metal roof rely on local physical tests and lacks the cyclic load sequence matching with regional climate, this paper proposed a method of constructing the fatigue load spectrum based on integration of wind load model, measured data of long-span metal roof and climate statistical data. According to the turbulence characteristics of wind, the wind load model is established from the aspects of turbulence intensity, power spectral density and wind pressure coefficient. Considering the influence of roof configuration on wind pressure distribution, the parameters are modified through fusing the measured data with least squares method to approximate the actual wind pressure load of the roof system. Furthermore, with regards to the wind climate characteristics of building location, Weibull model is adopted to analyze the regional meteorological data to obtain the probability density distribution of wind velocity used for calculating wind load, so as to establish the cyclic wind load sequence with the attributes of regional climate and building configuration. Finally, taking a workshop's metal roof as an example, the wind load spectrum is constructed according to this method, and the fatigue simulation and residual life prediction are implemented based on the experimental data. The forecasting result is lightly higher than the design standards, consistent with general principles of its conservative safety design scale, which shows that the presented method is validated for the fatigue characteristics study and health assessment of metal roof.

• 대한조선학회논문집
• /
• 제60권2호
• /
• pp.95-109
• /
• 2023

In the 4th industrial revolution, changes in the technological paradigm have had a direct impact on the maintenance system of ships. The 2-stroke low speed engine system integrates with the core equipment required for propulsive power. The Condition Based Management (CBM) is defined as a technology that predictive maintenance methods in existing calender-based or running time based maintenance systems by monitoring the condition of machinery and diagnosis/prognosis failures. In this study, we have established a framework for CBM technology development on our own, and are engaged in engineering-based failure analysis, data development and management, data feature analysis and pre-processing, and verified the reliability of failure mode DB using LSTM algorithms. We developed various simulated failure mode scenarios for 2-stroke low speed engine and researched to produce data on onshore basis test_beds. The analysis and pre-processing of normal and abnormal status data acquired through failure mode simulation experiment used various Exploratory Data Analysis (EDA) techniques to feature extract not only data on the performance and efficiency of 2-stroke low speed engine but also key feature data using multivariate statistical analysis. In addition, by developing an LSTM classification algorithm, we tried to verify the reliability of various failure mode data with time-series characteristics.

• 응용통계연구
• /
• 제36권2호
• /
• pp.129-139
• /
• 2023

변동성의 예측은 자산의 리스크에 대비하는 데에 중요한 역할을 하기때문에 필수적이다. 인공지능을 통하여 이러한 복잡한 특성을 지닌 변동성 예측을 시도하였는데 기존 시계열 예측에 적합하다 알려진 LSTM (1997)과 GRU (2014)은 기울기 소실로 인한 문제, 방대한 연산량의 문제, 그로 인한 메모리양의 문제 등이 존재하였다. 변동성 데이터는 비정상성(non-stationarity)과 정상성(stationarity)을 모두 가지고 있는 특성이 있으며, 자산 가격 하방 쇼크에 더 큰 폭으로 상승하는 비대칭성과 상당한 장기 기억성, 시장에 큰 사건이 발생할 때 기존의 값들에 비해 이상치라 할 수 있을 정도의 예측할 수 없는 큰 값이 발생하는 특성들이 존재한다. 이렇게 여러 가지 복잡한 특성들은 하나의 모형으로 구조화되기 어려워서 전통적인 방식의 모형으로는 변동성에 대한 예측력을 높이기 어려운 면이 있다. 이러한 문제를 해결하기 위해 1D CNN의 발전된 형태인 causal TCN (causal temporal convolutional network) 모형을 변동성 예측에 적용하고, 예측력을 최대화 할 수 있는 TCN 구조를 설계하고자 하였다. S&P 500, DJIA, Nasdaq 지수에 해당하는 변동성 지수 VIX, VXD, and VXN, 에 대하여 예측력 비교를 하였으며, TCN 모형이 RNN 계열의 모형보다도 전반적으로 예측력이 높음을 확인하였다.

• Journal of Platform Technology
• /
• 제11권4호
• /
• pp.62-70
• /
• 2023

이 논문은 농수산물 생산자가 도매시장에 상품을 출하하는 시기와 양을 결정하는 것을 돕기 위한 시스템을 구축하기 위한 일련의 과정 중 일부이다. 기존 농수산물 출하모델에서 사용하지 않은 품질변화효과를 모델링하고, 그 통계적 유의성을 확인한 후, 시스템에 도입하는 것이 이 연구의 목적이다. 이를 위해 연구자는 품질변화효과를 측정할 수 있는 네 가지 모델을 개발하였다. 시간이 지남에 따라 1) 품질이 일정하게 떨어지는 경우, 2) 품질이 처음에 급속히 떨어지다가 나중에는 천천히 떨어지는 경우, 3) 품질이 처음에 천천히 떨어지다가 나중에는 급속히 떨어지는 경우, 4) 품질이 낮았다가 시간이 흐른 후 높아지다가 다시 감소하는 경우를 모델링하였다. 각 모델의 품질변화효과가 출하량에 미지는 영향을 2014-2021년 사이에 가락도매시장에서 거래된 사과를 대상으로 실증분석 해 본 결과에 따르면 네 모델 모두 품질변화효과에 유의성을 발견하였다. 그리고 네 모델 간 설명력에 유의한 차이는 없었다. 따라서 네 개 모델 중 어느 하나를 선택해서 사과에 대한 출하시기의사결정시스템에 적용시킬 수 있는 것으로 나타났다.

• 응용통계연구
• /
• 제36권4호
• /
• pp.295-307
• /
• 2023

HAR 모형은 간단한 선형 모형으로 실현 변동성의 장기기억성을 비교적 잘 설명할 수 있어 널리 쓰이고 있다. 하지만, 실현 변동성은 조건부 이분산성, 레버리지 효과, 변동성 집중 등과 같은 복잡한 특징을 보이고 있기에 단순 HAR 모형을 확장할 필요가 있다. 따라서 본 연구는 조건부 이분산성을 설명하는 GARCH 모형에 임계값에 따라 계수가 달라지는 비선형 모형인 임계 HAR 모형(THAR-GARCH)을 제안하고 그 추정 방법 및 예측 성능에 대해서 살펴보고자 한다. 보다 구체적으로 오차항의 등분산 가정을 벗어났기 때문에 모형의 계수를 추정하기 위해서 반복적인 가중최소제곱추정법을 제안하고 모의실험을 통해 일치성을 보였다. 또한 전세계 21개의 주요 주가 지수의 실현 변동성에 대한 예측 오차를 비교함으로써 제안한 GARCH 오차를 가지는 임계 HAR 모형이 일반적으로 더 우수한 예측력을 보임을 확인하였다.

• Nuclear Engineering and Technology
• /
• 제55권9호
• /
• pp.3423-3440
• /
• 2023

In this work, a multivariate time-series machine learning meta-model is developed to predict the transient response of a typical nuclear power plant (NPP) undergoing a steam generator tube rupture (SGTR). The model employs Recurrent Neural Networks (RNNs), including the Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), and a hybrid CNN-LSTM model. To address the uncertainty inherent in such predictions, a Bayesian Neural Network (BNN) was implemented. The models were trained using a database generated by the Best Estimate Plus Uncertainty (BEPU) methodology; coupling the thermal hydraulics code, RELAP5/SCDAP/MOD3.4 to the statistical tool, DAKOTA, to predict the variation in system response under various operational and phenomenological uncertainties. The RNN models successfully captures the underlying characteristics of the data with reasonable accuracy, and the BNN-LSTM approach offers an additional layer of insight into the level of uncertainty associated with the predictions. The results demonstrate that LSTM outperforms GRU, while the hybrid CNN-LSTM model is computationally the most efficient. This study aims to gain a better understanding of the capabilities and limitations of machine learning models in the context of nuclear safety. By expanding the application of ML models to more severe accident scenarios, where operators are under extreme stress and prone to errors, ML models can provide valuable support and act as expert systems to assist in decision-making while minimizing the chances of human error.