• Wind and Structures
• /
• v.19 no.6
• /
• pp.623-647
• /
• 2014

In this paper, a probability distribution which is consistent with the observed phenomenon at the roof corner and, also on other portions of the roof, of a low-rise building is proposed. The model is consistent with the choice of probability density function suggested by the statistical thermodynamics of open systems and turbulence modelling in fluid mechanics. After presenting the justification based on physical phenomenon and based on statistical arguments, the fit of alpha-stable distribution for prediction of extreme negative wind pressure coefficients is explored. The predictions are compared with those actually observed during wind tunnel experiments (using wind tunnel experimental data obtained from the aerodynamic database of Tokyo Polytechnic University), and those predicted by using Gumbel minimum and Hermite polynomial model. The predictions are also compared with those estimated using a recently proposed non-parametric model in regions where stability criterion (in skewness-kurtosis space) is satisfied. From the comparisons, it is noted that the proposed model can be used to estimate the extreme peak negative wind pressure coefficients. The model has an advantage that it is consistent with the physical processes proposed in the literature for explaining large fluctuations at the roof corners.

• Structural Engineering and Mechanics
• /
• v.84 no.5
• /
• pp.651-664
• /
• 2022

A precise prediction of the ultimate bond strength between rebar and surrounding concrete plays a major role in structural design, as it effects the load-carrying capacity and serviceability of a member significantly. In the present study, Gaussian models are employed for modelling bond strength of ribbed steel bars embedded in concrete. Gaussian models offer a non-parametric method based on Bayesian framework which is powerful, versatile, robust and accurate. Five different Gaussian models are explored in this paper-Gaussian Process (GP), Variational Heteroscedastic Gaussian Process (VHGP), Warped Gaussian Process (WGP), Sparse Spectrum Gaussian Process (SSGP), and Twin Gaussian Process (TGP). The effectiveness of the models is also evaluated in comparison to the numerous design formulae provided by the codes. The predictions from the Gaussian models are found to be closer to the experiments than those predicted using the design equations provided in various codes. The sensitivity of the models to various parameters, input feature space and sampling is also presented. It is found that GP, VHGP and SSGP are effective in prediction of the bond strength. For large data set, GP, VHGP, WGP and TGP can be computationally expensive. In such cases, SSGP can be utilized.

• Geomechanics and Engineering
• /
• v.3 no.4
• /
• pp.285-290
• /
• 2011

This article employs Multivariate Adaptive Regression Spline (MARS) for determination of friction capacity of driven piles in clay. MARS is non-parametric adaptive regression procedure. Pile length, pile diameter, effective vertical stress, and undrained shear strength are considered as input of MARS and the output of MARS is friction capacity. The developed MARS gives an equation for determination of $f_s$ of driven piles in clay. The results of the developed MARS have been compared with the Artificial Neural Network. This study shows that the developed MARS is a robust model for prediction of $f_s$ of driven piles in clay.

• Proceedings of The Korean Society of Agricultural and Forest Meteorology Conference
• /
• 2001.06a
• /
• pp.101-104
• /
• 2001

작물 발육단계의 정확한 진단은 그 시기의 생리적 반응을 이해하고 정확하고 정밀한 생육관리를 위해서 절대적으로 필요한 요소이다. 지금까지 벼의 발육단계 예측을 위한 모델에는 GDD를 이용하는 방법(이, 1972), 한 단계의 발육을 완료하는데 걸리는 기간(t)과 이 기간중의 평균기온, 평균일장의 단순회귀 또는 중회귀를 구하는 방법(Gao et al, 1989; Yin et al, 1995; 임, 1982), 평균발육속도(1/t)를 이 기간중의 평균온도와 평균일장의 함수로 표현해서 이를 적산하여 1이 되었을 때를 발육완료일로 나타내는 방법(이, 1987; 신 등, 2000), 발육기간이 시계열자료를 모두 고려하여 함수를 이용하지 않는 non-parametric 방법(이, 1991) 등이 있다.(중략)

• Journal of the Korean Institute of Intelligent Systems
• /
• v.11 no.3
• /
• pp.231-240
• /
• 2001

Conventional expert systems has been criticized due to its lack of capability to adapt to the changing decision-making environments. In literature, many methods have been proposed to make expert systems more environment-adaptive by incorporating fuzzy logic and neural networks. The objective of this paper is to propose a new approach to building a self-evolving expert system inference mechanism by integrating fuzzy neural network and fuzzy rule extraction technique. The main recipe of our proposed approach is to fuzzify the training data, train them by a fuzzy neural network, extract a set of fuzzy rules from the trained network, organize a knowledge base, and refine the fuzzy rules by applying a pruning algorithm when the decision-making environments are detected to be changed significantly. To prove the validity, we tested our proposed self-evolving expert systems inference mechanism by using the bankruptcy data, and compared its results with the conventional neural network. Non-parametric statistical analysis of the experimental results showed that our proposed approach is valid significantly.

• Proceedings of the Korean Society for Technology of Plasticity Conference
• /
• 2009.05a
• /
• pp.425-428
• /
• 2009

Silicone is recently used for LED chip encapsulment due to its good thermal stability and optical transmittance. To mold a solid-state silicone encapsulment, curing by mixing at elevated temperatures followed by cooling is necessary. As the silicone molding process is involved in healing and subsequent cooling, the thermal residual stress, which causes mechanical warpage or optical birefringence in the final silicone encapsulment, may be induced if there are non-uniformities in cured silicone material properties or encapsulment shape design. The prediction of residual stress is necessary to design a high-quality silicone molding process. Therefore, in the present paper, a numerical parametric study was attempted to evaluate the heating and cooling effects on the thermal residual stress induced in the cured silicone.

• Journal of Industrial Technology
• /
• v.43 no.1
• /
• pp.57-62
• /
• 2023

In this review, we introduce the non-parametric Bayesian filtering algorithm known as the point-mass filter (PMF) and discuss recent studies related to it. PMF realizes Bayesian filtering by placing a deterministic grid on the state space and calculating the probability density at each grid point. PMF is known for its robustness and high accuracy compared to other nonparametric Bayesian filtering algorithms due to its uniform sampling. However, a drawback of PMF is its inherently high computational complexity in the prediction phase. In this review, we aim to understand the principles of the PMF algorithm and the reasons for the high computational complexity, and summarize recent research efforts to overcome this challenge. We hope that this review contributes to encouraging the consideration of PMF applications for various systems.

• Bulletin of the Korean Space Science Society
• /
• 2011.04a
• /
• pp.30.1-30.1
• /
• 2011

We present construction of 3D Earth optical Model for in-orbit performance prediction of L1 halo orbiting earth remote sensing instrument; the Albedo Monitor and Radiometer (Amon-Ra) using Integrated Ray Tracing (IRT) computational technique. The 3 components are defined in IRT; 1) Sun model, 2) Earth system model (Atmosphere, Land and Ocean), 3)Amon-Ra Instrument model. In this report, constructed sun model has Lambertian scattering hemisphere structure. The atmosphere is composed of 16 distributed structures and each optical model includes scatter model with both reflecting and transmitting direction respond to 5 deg. intervals of azimuth and zenith angles. Land structure model uses coastline and 5 kinds of vegetation distribution data structure, and its non-Lambertian scattering is defined with the semi-empirical "parametric kernel method" used for MODIS (NASA) missions. The ocean model includes sea ice cap with the sea ice area data from NOAA, and sea water optical model which is considering non-Lambertian sun-glint scattering. The IRT computation demonstrate that the designed Amon-Ra optical system satisfies the imaging and radiometric performance requirement. The technical details of the 3D Earth Model, IRT model construction and its computation results are presented together with future-works.

• Journal of Intelligence and Information Systems
• /
• v.28 no.2
• /
• pp.19-31
• /
• 2022

In Machine learning, we usually divide the entire data into training data and test data, train the model using training data, and use test data to determine the accuracy and generalization performance of the model. In the case of models with low generalization performance, the prediction accuracy of newly data is significantly reduced, and the model is said to be overfit. This study is about a method of generating training data based on central limit theorem and combining it with existed training data to increase normality and using this data to train models and increase generalization performance. To this, data were generated using sample mean and standard deviation for each feature of the data by utilizing the characteristic of central limit theorem, and new training data was constructed by combining them with existed training data. To determine the degree of increase in normality, the Kolmogorov-Smirnov normality test was conducted, and it was confirmed that the new training data showed increased normality compared to the existed data. Generalization performance was measured through differences in prediction accuracy for training data and test data. As a result of measuring the degree of increase in generalization performance by applying this to K-Nearest Neighbors (KNN), Logistic Regression, and Linear Discriminant Analysis (LDA), it was confirmed that generalization performance was improved for KNN, a non-parametric technique, and LDA, which assumes normality between model building.

• Earthquakes and Structures
• /
• v.7 no.4
• /
• pp.485-510
• /
• 2014

In performance-based seismic design procedures Peak Ground Acceleration (PGA) and pseudo-Spectral acceleration ($S_a$) are commonly used to predict the response of structures to earthquake. Recently, research has been carried out to evaluate the predictive capability of these standard Intensity Measures (IMs) with respect to different types of structures and Engineering Demand Parameter (EDP) commonly used to measure damage. Efforts have been also spent to propose alternative IMs that are able to improve the results of the response predictions. However, most of these IMs are not usually employed in probabilistic seismic demand analyses because of the lack of reliable Ground Motion Prediction Equations (GMPEs). In order to define seismic hazard and thus to calculate demand hazard curves it is essential, in fact, to establish a GMPE for the earthquake intensity. In the light of this need, new GMPEs are proposed here for the elastic input energy spectra, energy-based intensity measures that have been shown to be good predictors of both structural and non-structural damage for many types of structures. The proposed GMPEs are developed using mixed-effects models by empirical regressions on a large number of strong-motions selected from the NGA database. Parametric analyses are carried out to show the effect of some properties variation, such as fault mechanism, type of soil, earthquake magnitude and distance, on the considered IMs. Results of comparisons between the proposed GMPEs and other from the literature are finally shown.