• Nuclear Engineering and Technology
• /
• v.54 no.5
• /
• pp.1825-1834
• /
• 2022

Performing high-fidelity computational fluid dynamics (HF-CFD) to predict the flow and heat transfer state of the coolant in the reactor core is expensive, especially in scenarios that require extensive parameter search, such as uncertainty analysis and design optimization. This work investigated the performance of utilizing a multi-fidelity reduced-order model (MF-ROM) in PWR rod bundles simulation. Firstly, basis vectors and basis vector coefficients of high-fidelity and low-fidelity CFD results are extracted separately by the proper orthogonal decomposition (POD) approach. Secondly, a surrogate model is trained to map the relationship between the extracted coefficients from different fidelity results. In the prediction stage, the coefficients of the low-fidelity data under the new operating conditions are extracted by using the obtained POD basis vectors. Then, the trained surrogate model uses the low-fidelity coefficients to regress the high-fidelity coefficients. The predicted high-fidelity data is reconstructed from the product of extracted basis vectors and the regression coefficients. The effectiveness of the MF-ROM is evaluated on a flow and heat transfer problem in PWR fuel rod bundles. Two data-driven algorithms, the Kriging and artificial neural network (ANN), are trained as surrogate models for the MF-ROM to reconstruct the complex flow and heat transfer field downstream of the mixing vanes. The results show good agreements between the data reconstructed with the trained MF-ROM and the high-fidelity CFD simulation result, while the former only requires to taken the computational burden of low-fidelity simulation. The results also show that the performance of the ANN model is slightly better than the Kriging model when using a high number of POD basis vectors for regression. Moreover, the result presented in this paper demonstrates the suitability of the proposed MF-ROM for high-fidelity fixed value initialization to accelerate complex simulation.

• Korean Journal of Remote Sensing
• /
• v.28 no.1
• /
• pp.129-143
• /
• 2012

This paper demonstrates the effectiveness of advanced spatial modeling techniques for environmental monitoring and impact assessment through a case study of Chernobyl nuclear accident occurred in 1986. Land-cover types changed after the accident are analysed by a post classification comparison method using bi-temporal Landsat TM data acquired in 1986 and 1992 near the accident site. Spatial modeling including various kriging algorithms are also applied to analyze the relationships between Cesium concentrations in soil and thyroid cancer incidence rates in Belarus, which was greatly damaged by the accident. The change detection results clearly showed the decrease of croplands and the increase of abandoned lands, and concrete structures were newly built around the nuclear plant to prevent the spread of radioactive contamination. In Belarus, high Cesium concentrations were observed in southern areas with high thyroid cancer risk estimated by Poisson kriging. Geographically weighted regression, which could account for geographic variations of independent variables including Cesium concentrations and distances from the Chernobyl nuclear power plant, was applied to extract the relationships between the independent variables and the thyroid cancer risk. The estimated risk values showed a correlation coefficient value of 0.98 with respect to the thyroid cancer risk values, which implied that the thyroid cancer risk in Belarus was affected by the accident. In conclusion, it is expected that advanced spatial modeling techniques applied in this study would be useful for environmental impact assessment and public health research.

• Korean Journal of Remote Sensing
• /
• v.33 no.4
• /
• pp.445-454
• /
• 2017

The performance of spatial downscaling models depends on the quality of input coarse scale products. Thus, the impact of intrinsic errors contained in coarse scale satellite products on predictive performance should be properly assessed in parallel with the development of advanced downscaling models. Such an assessment is the main objective of this paper. Based on a synthetic satellite precipitation product at a coarse scale generated from rain gauge data, two synthetic precipitation products with different amounts of error were generated and used as inputs for spatial downscaling. Geographically weighted regression, which typically has very high explanatory power, was selected as the trend component estimation model, and area-to-point kriging was applied for residual correction in the spatial downscaling experiment. When errors in the coarse scale product were greater, the trend component estimates were much more susceptible to errors. But residual correction could reduce the impact of the erroneous trend component estimates, which improved the predictive performance. However, residual correction could not improve predictive performance significantly when substantial errors were contained in the input coarse scale data. Therefore, the development of advanced spatial downscaling models should be focused on correction of intrinsic errors in the coarse scale satellite product if a priori error information could be available, rather than on the application of advanced regression models with high explanatory power.

• Korean Journal of Remote Sensing
• /
• v.33 no.1
• /
• pp.25-35
• /
• 2017

Spatial downscaling with fine resolution auxiliary variables has been widely applied to predict precipitation at fine resolution from coarse resolution satellite-based precipitation products. The spatial downscaling framework is usually based on the decomposition of precipitation values into trend and residual components. The fine resolution auxiliary variables contribute to the estimation of the trend components. The main focus of this study is on quantitative analysis of impacts of trend component estimates on predictive performance in spatial downscaling. Two regression models were considered to estimate the trend components: multiple linear regression (MLR) and geographically weighted regression (GWR). After estimating the trend components using the two models,residual components were predicted at fine resolution grids using area-to-point kriging. Finally, the sum of the trend and residual components were considered as downscaling results. From the downscaling experiments with time-series Tropical Rainfall Measuring Mission (TRMM) 3B43 precipitation data, MLR-based downscaling showed the similar or even better predictive performance, compared with GWR-based downscaling with very high explanatory power. Despite very high explanatory power of GWR, the relationships quantified from TRMM precipitation data with errors and the auxiliary variables at coarse resolution may exaggerate the errors in the trend components at fine resolution. As a result, the errors attached to the trend estimates greatly affected the predictive performance. These results indicate that any regression model with high explanatory power does not always improve predictive performance due to intrinsic errors of the input coarse resolution data. Thus, it is suggested that the explanatory power of trend estimation models alone cannot be always used for the selection of an optimal model in spatial downscaling with fine resolution auxiliary variables.

• Korean Journal of Remote Sensing
• /
• v.37 no.6_1
• /
• pp.1739-1756
• /
• 2021

Atmospheric nitrogen dioxide (NO₂) is mainly caused by anthropogenic emissions. It contributes to the formation of secondary pollutants and ozone through chemical reactions, and adversely affects human health. Although ground stations to monitor NO₂ concentrations in real time are operated in Korea, they have a limitation that it is difficult to analyze the spatial distribution of NO₂ concentrations, especially over the areas with no stations. Therefore, this study conducted a comparative experiment of spatial interpolation of NO₂ concentrations based on two linear-regression methods(i.e., multi linear regression (MLR), and regression kriging (RK)), and two machine learning approaches (i.e., random forest (RF), and support vector regression (SVR)) for the year of 2020. Four approaches were compared using leave-one-out-cross validation (LOOCV). The daily LOOCV results showed that MLR, RK, and SVR produced the average daily index of agreement (IOA) of 0.57, which was higher than that of RF (0.50). The average daily normalized root mean square error of RK was 0.9483%, which was slightly lower than those of the other models. MLR, RK and SVR showed similar seasonal distribution patterns, and the dynamic range of the resultant NO₂ concentrations from these three models was similar while that from RF was relatively small. The multivariate linear regression approaches are expected to be a promising method for spatial interpolation of ground-level NO₂ concentrations and other parameters in urban areas.

• Journal of Urban Science
• /
• v.9 no.2
• /
• pp.31-44
• /
• 2020

This study aims to analyze the relations among greenbelt, air pollution empirically in order to assess the environmental effects of the greenbelt in the Seoul metropolitan area, objectively. For this purpose, this study conducts an empirical analysis of impacts of greenbelt on urban air pollution using a multiple-regression model. The major findings are summarized as follows. As a result of an empirical analysis of the impacts of greenbelt on air pollution, it is found that the characteristics of the city have impacts on air pollution concentration. It is found that the population and employment are the causes of increases in CO and NO2 concentrations, and the number of employees in the manufacturers has impacts on increases of O3 and SO2, while power plants have impacts on PM10, CO and NO2. Intersections have impacts on O3 and SO2, while the areas of the roads have impacts on CO and NO2. In addition, as for the spatial distribution of air pollutants, it is found that CO and NO2 concentrations are relatively higher in the center of the Seoul metropolitan area, while PM10, O3 and SO2 concentrations are relatively higher in the suburbs. It is found that air pollution concentration is low in greenbelt zone. In the greenbelt zone, PM10, CO and SO2 concentrations are low.

• Smart Structures and Systems
• /
• v.32 no.1
• /
• pp.61-81
• /
• 2023

This study presents an ensemble learning based Bayesian model updating approach for structural damage diagnosis. In the developed framework, the structure is initially decomposed into a set of substructures. The autoregressive moving average (ARMAX) model is established first for structural damage localization based structural motion equation. The wavelet packet decomposition is utilized to extract the damage-sensitive node energy in different frequency bands for constructing structural surrogate models. Four methods, including Kriging predictor (KRG), radial basis function neural network (RBFNN), support vector regression (SVR), and multivariate adaptive regression splines (MARS), are selected as candidate structural surrogate models. These models are then resampled by bootstrapping and combined to obtain an ensemble model by probabilistic ensemble. Meanwhile, the maximum entropy principal is adopted to search for new design points for sample space updating, yielding a more robust ensemble model. Through the iterations, a framework of surrogate ensemble learning based model updating with high model construction efficiency and accuracy is proposed. The specificities of the method are discussed and investigated in a case study.

• Proceedings of the KSME Conference
• /
• 2008.11a
• /
• pp.1314-1319
• /
• 2008

Fatigue reliability problems are nowadays actively considered in the design of mechanical components. Recently, Dimension Reduction Method using Kriging approximation (KDRM) was proposed by the authors to efficiently calculate statistical moments of the response function. This method, which is more tractable for its sensitivity-free nature and providing the response PDF in a few number of analyses, is adopted in this study for the reliability analysis. Before applying this method to the practical fatigue problems, accuracies are studied in terms of parameters of the KDRM through a number of numerical examples, from which best set of parameters are suggested. In the fatigue reliability problems, good number of experimental data are necessary to get the statistical distribution of the S-N parameters. The information, however, are not always available due to the limited expense and time. In this case, a family of curves with prediction interval, called P-S-N curve, is constructed from regression analysis. Using the KDRM, once a set of responses are available at the sample points at the mean, all the reliability analyses for each P-S-N curve can be efficiently studied without additional response evaluations. The method is applied to a spring design problem as an illustration of practical applications, in which reliability-based design optimization (RBDO) is conducted by employing stochastic response surface method which includes probabilistic constraints in itself. Resulting information is of great practical value and will be very helpful for making trade-off decision during the fatigue design.

• The Korean Journal of Applied Statistics
• /
• v.26 no.1
• /
• pp.93-109
• /
• 2013

Geostatistical data or point-referenced data have the information on the monitoring stations of interest where the observations are measured. Practical geostatistical data are obtained from a wide variety of observational monitoring networks that are mainly operated by the Korean government. When we analyze geostatistical data and predict the expectations at unobservable locations, we can improve the reliability of the prediction by utilizing some relevant spatial data obtained from different observational monitoring networks and blend them with the measurements of our main interest. In this paper, we consider the hierarchical spatial linear model that enables us to link spatial variables from different resources but with similar patterns and guarantee the precision of the prediction. We compare the proposed model to a classical linear regression model and simple kriging in terms of some information criteria and one-leave-out cross-validation. Real application deals with Sulfur Dioxide($SO_2$) measurements from the urban air pollution monitoring network and wind speed data from the surface observation network.

• Water Engineering Research
• /
• v.6 no.2
• /
• pp.39-48
• /
• 2005

The Brazos River is one of the longest rivers contained entirely in the state of Texas, flowing over 700 miles from northwest Texas to the Gulf of Mexico. Today, the Brazos River Authority and Texas Commission on Environmental Quality interest in drought protection plan, waterpower project, and allowing the appropriation of water system-wide and water right within the Brazos River Basin to meet water needs of customers like farmers and local civilians in the future. Especially, this purpose of this paper primarily intended to provide the data for the engineering guidelines and make easily geological mapping tool. In the Brazos River basin, many stream-flow gage station sites are not working, and they can not provide stream-flow data sets enough for development of the Probable Maximum Flood (PMF) for use in the evaluation of proposed and existing dams and other impounding structures. Integrated GIS-Spatial Interpolation (GIS-SI) tool are composed of two parts; (1) extended GIS technique (new making interface for hydrological regionalization parameters plus classical GIS mapping skills), (2) Spatial Interpolation technique using weighting factors from kriging method. They are obtained from the relationship among location and elevation of geological watershed and existing stream-flow datasets. GIS-SI technique is easily used to compute parameters which get drainage areas, mean daily/monthly/annual precipitation, and weighted values. Also, they are independent variables of multiple linear regressions for simulation at un gaged stream-flow sites. In this study, GIS-SI technique is applied to the Brazos river basin in Texas. By assuming the ungaged flow at the sites of Palo Pinto, Bryan and Needville, the simulated daily/monthly/annual time series are compared with observed time series. The simulated daily/monthly/annual time series are highly correlated with and well fitted to the observed times series.