• Proceedings of the Korean Society of Disaster Information Conference
• /
• 2023.11a
• /
• pp.303-304
• /
• 2023

확률론적 위험성 평가는 하중, 재료특성 등과 같은 불확실성 인자를 고려하여 구조물의 안전성을 평가하는 기법이지만 모든 불확실성을 고려하는 것은 현실적으로 불가능하다. 또한 원전 격납건물은 콘크리트, 철근, 라이너, 텐던이 복잡하게 결합되어 있다. 따라서 전역민감도 분석을 통해 격납건물의 불확실성 인자 검토하고 선정하는 작업은 필요하다. 따라서 본 연구는 대리모델을 기반으로 축소규모 원전 격납건물의 전역 민감도 분석을 수행하고 격납건물의 주요 영향인자를 분석하고자 한다. 유한요소 해석 모델을 기반으로 대리모델의 학습데이터를 생성하였으며 구축된 대리모델의 성능지표를 분석하였을 때 높은 회귀성능을 갖는 것으로 판단된다. 대리모델을 기반으로 전역 민감도 분석을 수행한 결과 콘크리트의 인장균열이 발생하는 내압수준에서 민감도 지수는 콘크리트의 압축강도가 높지만, 전체적인 내압 구간에서 민감도 지수는 텐던의 탄성계수 및 항복강도가 높은 것으로 나타났다.

• Korean Chemical Engineering Research
• /
• v.61 no.4
• /
• pp.542-549
• /
• 2023

Since the growing interest in surrogate modeling, there has been continuous research aimed at simulating nonlinear chemical processes using data-driven machine learning. However, the opaque nature of machine learning models, which limits their interpretability, poses a challenge for their practical application in industry. Therefore, this study aims to analyze chemical processes using Explainable Artificial Intelligence (XAI), a concept that improves interpretability while ensuring model accuracy. While conventional sensitivity analysis of chemical processes has been limited to calculating and ranking the sensitivity indices of variables, we propose a methodology that utilizes XAI to not only perform global and local sensitivity analysis, but also examine the interactions among variables to gain physical insights from the data. For the ammonia synthesis process, which is the target process of the case study, we set the temperature of the preheater leading to the first reactor and the split ratio of the cold shot to the three reactors as process variables. By integrating Matlab and Aspen Plus, we obtained data on ammonia production and the maximum temperatures of the three reactors while systematically varying the process variables. We then trained tree-based models and performed sensitivity analysis using the SHAP technique, one of the XAI methods, on the most accurate model. The global sensitivity analysis showed that the preheater temperature had the greatest effect, and the local sensitivity analysis provided insights for defining the ranges of process variables to improve productivity and prevent overheating. By constructing alternative models for chemical processes and using XAI for sensitivity analysis, this work contributes to providing both quantitative and qualitative feedback for process optimization.

• Journal of the Society of Disaster Information
• /
• v.19 no.4
• /
• pp.976-983
• /
• 2023

Purpose: The dynamic behavior of a bridge structure under seismic loading depends on many uncertainties, such as the nature of the seismic waves and the material and geometric properties. However, not all uncertainties have a significant impact on the dynamic behavior of a bridge structure. Since probabilistic seismic performance evaluation considering even low-impact uncertainties is computationally expensive, the uncertainties should be identified by considering their impact on the dynamic behavior of the bridge. Therefore, in this study, a global sensitivity analysis was performed to identify the main parameters affecting the dynamic behavior of bridges with I-curved girders. Method: Considering the uncertainty of the earthquake and the material and geometric uncertainty of the curved bridge, a finite element analysis was performed, and a surrogate model was developed based on the analysis results. The surrogate model was evaluated using performance metrics such as coefficient of determination, and finally, a global sensitivity analysis based on the surrogate model was performed. Result: The uncertainty factors that have the greatest influence on the stress response of the I-curved girder under seismic loading are the peak ground acceleration (PGA), the height of the bridge (h), and the yield stress of the steel (fy). The main effect sensitivity indices of PGA, h, and fy were found to be 0.7096, 0.0839, and 0.0352, respectively, and the total sensitivity indices were found to be 0.9459, 0.1297, and 0.0678, respectively. Conclusion: The stress response of the I-shaped curved girder is dominated by the uncertainty of the input motions and is strongly influenced by the interaction effect between each uncertainty factor. Therefore, additional sensitivity analysis of the uncertainty of the input motions, such as the number of input motions and the intensity measure(IM), and a global sensitivity analysis considering the structural uncertainty, such as the number and curvature of the curved girders, are required.

• Journal of Korea Water Resources Association
• /
• v.31 no.6
• /
• pp.695-708
• /
• 1998

Recently, it seems increasing trends that the design floods on the middle and small scale of urban regions and natural basins are evaluated with introducing to the concept of the critical storm duration. However the study of the critical storm duration is not sufficient and especially on the natural basins, it rarely performed. therefore in this study, estimated the critical storm duration and peak discharge according to the rainfall distribution type, the position of peak rainfall intensity, and the frequency on the natural basins were evaluated using Clark model and the influence of each factors on the design floods was analyzed with sensitivity analysis on the parameters of the model.

• Journal of Korea Water Resources Association
• /
• v.39 no.8 s.169
• /
• pp.677-690
• /
• 2006

The LH-OAT (Latin Hypercube One factor At a Time) method for sensitivity analysis and SCE-UA (Shuffled Complex Evolution at University of Arizona) optimization method were applied for the automatic calibration of SWAT model in Bocheong-cheon watershed. The LH-OAT method which combines the advantages of global and local sensitivity analysis effectively identified the sensitivity ranking for the parameters of SWAT model over feasible parameter space. Use of this information allows us to select the calibrated parameters for the automatic calibration process. The performance of the automatic calibration of SWAT model using SCE-UA method depends on the length of calibration period, the number of calibrated parameters, and the selection of statistical error criteria. The performance of SWAT model in terms of RMSE (Root Mean Square Error), NSEF (Nash-Sutcliffe Model Efficiency), RMAE (Relative Mean Absolute Error), and NMSE (Normalized Mean Square Error) becomes better as the calibration period and the number of parameters defined in the automatic calibration process increase. However, NAE (Normalized Average Error) and SDR (Standard Deviation Ratio) were not improved although the calibration period and the number of calibrated parameters are increased. The result suggests that there are complex interactions among the calibration data, the calibrated parameters, and the model error criteria and a need for further study to understand these complex interactions at various representative watersheds.

• Journal of Digital Contents Society
• /
• v.11 no.1
• /
• pp.17-24
• /
• 2010

The purpose of this study is to design and implement a kernel for the support vector machine(SVM) to improve the performance of face recognition. Local feature analysis(LFA) has been well known for its good performance. SVM kernel plays a limited role of mapping low dimensional face features to high dimensional feature space but the proposed kernel using LFA is designed for face recognition purpose. Because of the novel method that local face information is extracted from training set and combined into the kernel, this method is expected to apply to various object recognition/detection tasks. The experimental results shows its improved performance.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2016.05a
• /
• pp.529-533
• /
• 2016

미국 환경청의 SWMM 5(storm water management model 5)는 유역의 홍수유출 모의 및 연속 유출 모의를 할 수 있는 모형으로서 국내뿐만 아니라 세계적으로 많이 사용되고 있는 모형이다. SWMM 5와 같은 유역 유출모형에서 결과의 적절성을 향상시키기 위해서는 모형에 사용되는 매개변수를 올바르게 추정할 필요가 있다. 하지만, 외국의 정교한 유역 유출모형들이 우리나라에서 제대로 적용되고 있지 못하는 이유 중 하나는 적절한 매개변수의 추정이 이루어지지 못하고 있는 점이다. 이러한 문제를 해결하고자 SWMM 5의 매개변수 추정 지원 시스템을 개발하였다. SWMM 5의 매개변수 추정지원 시스템은 민감도 분석, 최적화 기법에 의한 모형 자동보정, 매개변수 할당 및 도움 모듈로 이루어져 있다. SWMM 5의 매개변수 추정 지원 시스템에 사용되는 최적화 기법은 전역최적화 기법 중 하나인 SCE-UA(shuffled complex evolution-University of Arizona) 이다. SWMM 5의 매개변수 추정 지원 시스템의 개발은 국내 수자원 기술자들의 SWMM 5에 대한 이해 및 활용도를 더욱 향상시켜줄 것으로 기대한다.

• KSCE Journal of Civil and Environmental Engineering Research
• /
• v.43 no.3
• /
• pp.337-352
• /
• 2023

In general, Rainfall-Runoff model parameter set is optimized using the entire data to calculate unique parameter set. However, Korea has a large precipitation deviation according to the season, and it is expected to even worsen due to climate change. Therefore, the need for hydrological data considering seasonal characteristics. In this study, we conducted regional sensitivity analysis(RSA) using the conceptual Rainfall-Runoff model, GR4J aimed at the Soyanggang dam basin, and clustered combining the RSA results with hydrometeorological data using Self-Organizing map(SOM). In order to consider the climate characteristics in parameter estimation, the data was divided based on clustering, and a calibration approach of the Rainfall-Runoff model was developed by comparing the objective functions of the Global Optimization method. The performance of calibration was evaluated by statistical techniques. As a result, it was confirmed that the model performance during the Cold period(November~April) with a relatively low flow rate was improved. This is expected to improve the performance and predictability of the hydrological model for areas that have a large precipitation deviation such as Monsoon climate.

• Journal of the Architectural Institute of Korea Structure & Construction
• /
• v.36 no.5
• /
• pp.177-185
• /
• 2020

The building energy performance indicator, called Energy Performance Index (EPI), has been used for the past decades in South Korea. It has a list of design variables assigned with weighting factors (a, b). Unfortunately, the current EPI method is not performance-based but very close to a prescriptive rating. With this in mind, this study aims to propose a new performance-based EPI method. For this purpose, a global sensitivity analysis method, Sobol, is employed. The Sobol method is suitable for complex nonlinear models and can decompose all the output variance due to every input. The Sobol sensitivity index of each variable is defined as 0 to 1 (0 to 100%), and the sum of all sensitivity indices is equal to 1 (100%). In this study, an office building was modeled using EnergyPlus and then the Latin Hypercube Sampling (LHS) was conducted to generate a surrogate model to EnergyPlus. The sensitivity index was suggested to replace weight (a) in the existing EPI. In addition, the discrete weight (b) in the existing EPI was replaced by a set of continuous regression functions. Due to the introduction of the sensitivity index and the continuous regression functions, the new proposed approach can provide far more accurate outcome than the existing EPI (R²: 0.83 vs. R²: 0.01 for cooling, R²: 0.66 vs. R²: 0.01 for total energy). The new proposed approach proves to be more rational, objective and performance-based than the existing EPI method.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2024.05a
• /
• pp.749-752
• /
• 2024

연합학습은 클라이언트가 중앙 서버에 원본 데이터를 주지 않고도 학습할 수 있도록 설계된 분산된 머신러닝 방법이다. 그러나 클라이언트와 중앙 서버 사이에 모델 업데이트 정보를 공유한다는 점에서 여전히 추론 공격(Inference Attack)과 오염 공격(Poisoning Attack)의 위험에 노출되어 있다. 이러한 공격을 방어하기 위해 연합학습에 차분프라이버시(Differential Privacy)를 적용하는 방안이 연구되고 있다. 차분 프라이버시는 데이터에 노이즈를 추가하여 민감한 정보를 보호하면서도 유의미한 통계적 정보 쿼리는 공유할 수 있도록 하는 기법으로, 노이즈를 추가하는 위치에 따라 전역적 차분프라이버시(Global Differential Privacy)와 국소적 차분 프라이버시(Local Differential Privacy)로 나뉜다. 이에 본 논문에서는 차분 프라이버시를 적용한 연합학습의 최신 연구 동향을 전역적 차분 프라이버시를 적용한 방향과 국소적 차분 프라이버시를 적용한 방향으로 나누어 검토한다. 또한 이를 세분화하여 차분 프라이버시를 발전시킨 방식인 적응형 차분 프라이버시(Adaptive Differential Privacy)와 개인화된 차분 프라이버시(Personalized Differential Privacy)를 응용하여 연합학습에 적용한 방식들에 대하여 특징과 장점 및 한계점을 분석하고 향후 연구방향을 제안한다.