• 한국지능시스템학회:학술대회논문집
• /
• 한국퍼지및지능시스템학회 1998년도 추계학술대회 학술발표 논문집
• /
• pp.232-235
• /
• 1998

This paper evaluates the performance of a rough set based pattern classifier using the benchmarks in artificial neural nets depository found in internet. The definition of rough set in soft computing paradigm is briefly introduced. next the design of rough set classifier is suggested. Finally benchmark test results are shown the performance of rough set compare to that of ANNs and decision tree.

• Computers and Concrete
• /
• 제19권6호
• /
• pp.711-716
• /
• 2017

This paper presents a method using artificial neural networks (ANNs) to predict the residual moment capacity of thermally insulated reinforced concrete (RC) beams exposed to fire. The use of heat resistant insulation material protects concrete beams against the harmful effects of fire. If it is desired to calculate the residual moment capacity of the beams in this state, the determination of the moment capacity of thermally insulated beams exposed to fire involves several consecutive calculations, which is significantly easier when ANNs are used. Beam width, beam effective depth, fire duration, concrete compressive and steel tensile strength, steel area, thermal conductivity of insulation material can influence behavior of RC beams exposed to high temperatures. In this study, a finite difference method was used to calculate the temperature distribution in a cross section of the beam, and temperature distribution, reduction mechanical properties of concrete and reinforcing steel and moment capacity were calculated using existing relations in literature. Data was generated for 336 beams with different beam width ($b_w$), beam account height (h), fire duration (t), mechanical properties of concrete ($f_{cd}$) and reinforcing steel ($f_{yd}$), steel area ($A_s$), insulation material thermal conductivity (kinsulation). Five input parameters ($b_w$, h, $f_{cd}$, $f_{yd}$, $A_s$ and $k_{insulation}$) were used in the ANN to estimate the moment capacity ($M_r$). The trained model allowed the investigation of the effects on the moment capacity of the insulation material and the results indicated that the use of insulation materials with the smallest value of the thermal conductivities used in calculations is effective in protecting the RC beam against fire.

• Computers and Concrete
• /
• 제2권1호
• /
• pp.79-96
• /
• 2005

The use of high-strength concrete (HSC) has significantly increased over the last decade, especially in offshore structures, long-span bridges, and tall buildings. The behavior of such concrete is noticeably different from that of normal-strength concrete (NSC) due to its different microstructure and mode of failure. In particular, the shear capacity of structural members made of HSC is a concern and must be carefully evaluated. The shear fracture surface in HSC members is usually trans-granular (propagates across coarse aggregates) and is therefore smoother than that in NSC members, which reduces the effect of shear transfer mechanisms through aggregate interlock across cracks, thus reducing the ultimate shear strength. Current code provisions for shear design are mainly based on experimental results obtained on NSC members having compressive strength of up to 50MPa. The validity of such methods to calculate the shear strength of HSC members is still questionable. In this study, a new approach based on artificial neural networks (ANNs) was used to predict the shear capacity of NSC and HSC beams without shear reinforcement. Shear capacities predicted by the ANN model were compared to those of five other methods commonly used in shear investigations: the ACI method, the CSA simplified method, Response 2000, Eurocode-2, and Zsutty's method. A sensitivity analysis was conducted to evaluate the ability of ANNs to capture the effect of main shear design parameters (concrete compressive strength, amount of longitudinal reinforcement, beam size, and shear span to depth ratio) on the shear capacity of reinforced NSC and HSC beams. It was found that the ANN model outperformed all other considered methods, providing more accurate results of shear capacity, and better capturing the effect of basic shear design parameters. Therefore, it offers an efficient alternative to evaluate the shear capacity of NSC and HSC members without stirrups.

• 한국정보과학회논문지:소프트웨어및응용
• /
• 제29권1_2호
• /
• pp.1-9
• /
• 2002

진화 신경망은 기존의 경험적 지식 대신에 진화 알고리즘의 전역 탐색 능력을 사용해서 최적의 신경망을 찾는다. 하지만 실세계의 복잡한 문제는 하나의 신경망으로 해결하기 어려운 경우가 많기 때문에 최근에 하나 이상의 신경망을 결합한 다중 신경망에 관한 연구가 활발히 진행되고 있다. 본 논문에서는 진화과정 중 상호보완 가능한 다양한 신경망을 얻기 위한 종분화 방식을 제안한다. 또한 적합도 공유를 통해 종분화된 진화 신경망의 결과를 효과적으로 결합하기 위해 추상 레벨, 순위 레벨, 측정치 레벨의 여러 결합 방법을 이용한 다중 신경망 시스템을 개발한다. UCI 데이터베이스의 벤치마크 문제 중 호주 신용카드 승인 데이터에 대하여 실험한 결과, 종분화를 사용해 탐색한 신경망을 결합한 경우는 더 높은 인식률을 보였으며 Borda 결합의 경우 0.105의 오류율을 보여 제안한 방법이 효과적임을 알 수 있었다.

• Advances in concrete construction
• /
• 제10권6호
• /
• pp.479-488
• /
• 2020

The experimental and numerical works were carried out on high performance fiber reinforced concrete (HPFRC) with w/cm ratios ranging from 0.25 to 0.40, fiber volume fraction (Vf)=0-1.5% and 10% silica fume replacement. Improvements in compressive and flexural strengths obtained for HPFRC are moderate and significant, respectively, Empirical equations developed for the compressive strength and flexural strength of HPFRC as a function of fiber volume fraction. A relation between flexural strength and compressive strength of HPFRC with R=0.78 was developed. Due to the complex mix proportions and non-linear relationship between the mix proportions and properties, models with reliable predictive capabilities are not developed and also research on HPFRC was empirical. In this paper due to the inadequacy of present method, a back propagation-neural network (BP-NN) was employed to estimate the 28-day compressive strength of HPFRC mixes. BP-NN model was built to implement the highly non-linear relationship between the mix proportions and their properties. This paper describes the data sets collected, training of ANNs and comparison of the experimental results obtained for various mixtures. On statistical analyses of collected data, a multiple linear regression (MLR) model with R2=0.78 was developed for the prediction of compressive strength of HPFRC mixes, and average absolute error (AAE) obtained is 6.5%. On validation of the data sets by NNs, the error range was within 2% of the actual values. ANN model has given the significant degree of accuracy and reliability compared to the MLR model. ANN approach can be effectively used to estimate the 28-day compressive strength of fibrous concrete mixes and is practical.

• Wind and Structures
• /
• 제36권6호
• /
• pp.393-404
• /
• 2023

Skewness and kurtosis are important higher-order statistics for simulating non-Gaussian wind pressure series on low-rise buildings, but their predictions are less studied in comparison with those of the low order statistics as mean and rms. The distribution gradients of skewness and kurtosis on roofs are evidently higher than those of mean and rms, which increases their prediction difficulty. The conventional artificial neural networks (ANNs) used for predicting mean and rms show unsatisfactory accuracy in predicting skewness and kurtosis owing to the limited capacity of shallow learning of ANNs. In this work, the deep neural networks (DNNs) model with the ability of deep learning is introduced to predict the skewness and kurtosis on a low-rise building. For obtaining the optimal generalization of the DNNs model, the hyper parameters are automatically determined by Bayesian Optimization (BO). Moreover, for providing a benchmark for future studies on predicting higher order statistics, the data sets for training and testing the DNNs model are extracted from the internationally open NIST-UWO database, and the prediction errors of all taps are comprehensively quantified by various error metrices. The results show that the prediction accuracy in this study is apparently better than that in the literature, since the correlation coefficient between the predicted and experimental results is 0.99 and 0.75 in this paper and the literature respectively. In the untrained cornering wind direction, the distributions of skewness and kurtosis are well captured by DNNs on the whole building including the roof corner with strong non-normality, and the correlation coefficients between the predicted and experimental results are 0.99 and 0.95 for skewness and kurtosis respectively.

• Computers and Concrete
• /
• 제34권1호
• /
• pp.93-122
• /
• 2024

One well-known reason for using Fiber Reinforced Polymer (FRP) composites is to improve concrete strength and strain capacity via external confinement. Hence, various studies have been undertaken to offer a good illustration of the response of FRP-wrapped concrete for practical design intents. However, in such studies, the strength and strain of the confined concrete were predicted using regression analysis based on a limited number of test data. This study presents an approach based on artificial neural networks (ANNs) to develop models to predict the strength and strain at maximum stress enhancement of circular concrete cross-sections confined with different FRP types (Carbone, Glass, Aramid). To achieve this goal, a large test database comprising 493 axial compression experiments on FRP-confined concrete samples was compiled based on an extensive review of the published literature and used to validate the predicted artificial intelligence techniques. The ANN approach is currently thought to be the preferred learning technique because of its strong prediction effectiveness, interpretability, adaptability, and generalization. The accuracy of the developed ANN model for predicting the behavior of FRP-confined concrete is commensurate with the experimental database compiled from published literature. Statistical measures values, which indicate a better fit, were observed in all of the ANN models. Therefore, compared to existing models, it should be highlighted that the newly developed models based on FRP type are remarkably accurate.

• 대한교통학회지
• /
• 제22권7호
• /
• pp.155-159
• /
• 2004

최근 교통량 예측을 위한 인공 신경망(Artificial neural networks : ANNs) 구조와 학습방법에 대한 연구가 다양하게 시도되고 있다. 이것은 신경망이 유연한 비선형 모형(non-linear model)으로 강력한 패턴 인식 능력을 가지고 있기 때문이다. 그러나, 신경망은 비선형 모형이기 때문에 많은 매개변수(parameter)를 사용하게 되면서 과적합(overfitting) 문제에 부딪히게 된다. 본 논문에서는 이러한 교통량 예측을 위한 신경망 모형에서 과적합을 해소하기 위한 방안으로 매개변수에 대한 다양한 모형선택기준(model selection criterion)에 대한 적용성에 대해서 알아보았다. 특히, AIC계열을 중심으로 모형선택기준으로 선택된 모형이 과적합 경향을 해소하고 시간적 전이성을 보장할 수 있는지를 분석하는데 본 연구의 목적을 두고 있다. 교통량 자료를 신경망 모형에 적용하여 분석한 결과, 첫째 학습자료(in-sample) 모형선택기준에 의해 선택된 모형이 검증자료(out-of-sample)의 최적의 성능을 보장하지는 못한다는 결과를 얻었다. 즉, 본 연구에서 기존의 연구에서처럼, 학습자료(in-sample)의 최적 모형이 검증자료(out-of-sample)의 성능과 직접적인 관계가 없다는 것을 알 수 있었다. 둘째 모형선택기준의 안정성을 분석한 결과 AIC3, AICC, BIC는 안정적인 모형을 선택하는 기준으로서 의미가 있는 것으로 분석되었다. 하지만, AIC4의 경우는 최상의 모형과 편차가 큰 것으로 분석되었다. 시계열 자료 분석과 예측에 있어서 모형의 불확실성은 학습 자료와 검증 자료의 상관관계에 영향을 줄 수 있음에 비춰볼 때, 앞으로 보다 많은 자료에 대한 분석이 필요하다고 판단되며, 다른 시계열 자료에 대한 분석이 요구된다. 수 없었지만, 확정적 통행배정모형으로 설정한 경우, Stackelberg게임 접근법이 Cournot-Nash게임 접근법 보다 더 우수함을 확인할 수 있었다.다.수안보 등 지역에서 나타난다 이러한 이상대 주변에는 대개 온천이 발달되어 있었거나 새로 개발되어 있는 곳이다. 온천에 이용하고 있는 시추공의 자료는 배제하였으나 온천이응으로 직접적으로 영향을 받지 않은 시추공의 자료는 사용하였다 이러한 온천 주변 지역이라 하더라도 실제는 온천의 pumping 으로 인한 대류현상으로 주변 일대의 온도를 올려놓았기 때문에 비교적 높은 지열류량 값을 보인다. 한편 한반도 남동부 일대는 이번 추가된 자료에 의해 새로운 지열류량 분포 변화가 나타났다 강원 북부 오색온천지역 부근에서 높은 지열류량 분포를 보이며 또한 우리나라 대단층 중의 하나인 양산단층과 같은 방향으로 발달한 밀양단층, 모량단층, 동래단층 등 주변부로 NNE-SSW 방향의 지열류량 이상대가 발달한다. 이것으로 볼 때 지열류량은 지질구조와 무관하지 않음을 파악할 수 있다. 특히 이러한 단층대 주변은 지열수의 순환이 깊은 심도까지 가능하므로 이러한 대류현상으로 지표부근까지 높은 지온 전달이 되어 나타나는 것으로 판단된다.의 안정된 방사성표지효율을 보였다. $^{99m}Tc$-transferrin을 이용한 감염영상을 성공적으로 얻을 수 있었으며, $^{67}Ga$-citrate 영상과 비교하여 더 빠른 시간 안에 우수한 영상을 얻을 수 있었다. 그러므로 $^{99m}Tc$-transierrin이 감염 병소의 영상진단에 사용될 수 있을 것으로 기대된다.리를 정량화 하였다. 특히 선조체에서의 도파민 유리에 의한 수용체 결합능의 감소는 흡연에 의한 혈중 니코틴의 축적 농도와 양의 상관관계를 보였다

• 한국정밀공학회지
• /
• 제17권6호
• /
• pp.176-184
• /
• 2000

Reverse engineering (RE) is a process to create computer aided design (CAD) models from the scanned data of an existing part acquired using 3D position scanners. This paper proposes a novel methodology of extracting geometric features directly from a set of 3D scanned points, which utilizes the concepts of feature-based technology and artificial neural networks (ANNs). The use of ANN has enabled the development of a flexible feature-based RE application that can be trained to deal with various features. The following four main tasks were mainly investigated and implemented: (1) Data reduction; (2) edge detection; (3) ANN-based feature recognition; (4) feature extraction. This approach was validated with a variety of real industrial components. The test results show that the developed feature-based RE application proved to be suitable for reconstructing prismatic features such as block, pocket, step, slot, hole, and boss, which are very common and crucial in mechanical engineering products.

• Journal of Electrical Engineering and Technology
• /
• 제6권6호
• /
• pp.760-768
• /
• 2011

The performance of transmission lines and its shielding design during a lightning phenomenon are quite essential in the maintenance of a reliable power supply to consumers. The leader progression model, as an advanced approach, has been recently developed to calculate the shielding failure rate (SFR) of transmission lines using geometrical data and physical behavior of upward and downward lightning leaders. However, such method is quite time consuming. In the present paper, an effective method that utilizes artificial neural networks (ANNs) to create a metamodel for calculating the SFR of a transmission line based on shielding angle and height is introduced. The results of investigations on a real case study reveal that, through proper selection of an ANN structure and good training, the ANN prediction is very close to the result of the detailed simulation, whereas the Processing time is by far lower than that of the detailed model.