• Structural Engineering and Mechanics
• /
• 제74권1호
• /
• pp.55-67
• /
• 2020

The common practice to predict the characteristic structural load effects (LEs) in long reference periods is to employ the extreme value theory (EVT) for building limit distributions. However, most applications ignore that LEs are driven by multiple loading events and thus do not have the identical distribution, a prerequisite for EVT. In this study, we propose the composite extreme value modeling approach using clustering to (a) cluster initial blended samples into finite identical distributed subsamples using the finite mixture model, expectation-maximization algorithm, and the Akaike information criterion; (b) combine limit distributions of subsamples into a composite prediction equation using the generalized Pareto distribution based on a joint threshold. The proposed approach was validated both through numerical examples with known solutions and engineering applications of bridge traffic LEs on a long-span bridge. The results indicate that a joint threshold largely benefits the composite extreme value modeling, many appropriate tail approaching models can be used, and the equation form is simply the sum of the weighted models. In numerical examples, the proposed approach using clustering generated accurate extrema prediction of any reference period compared with the known solutions, whereas the common practice of employing EVT without clustering on the mixture data showed large deviations. Real-world bridge traffic LEs are driven by multi-events and present multipeak distributions, and the proposed approach is more capable of capturing the tendency of tailed LEs than the conventional approach. The proposed approach is expected to have wide applications to general problems such as samples that are driven by multiple events and that do not have the identical distribution.

• The Journal of Advanced Prosthodontics
• /
• 제13권1호
• /
• pp.12-23
• /
• 2021

Purpose. The aim of the study was to compare the lingualized implant placement creating a buccal cantilever with prosthetic-driven implant placement exhibiting excessive crown-to-implant ratio. Materials and Methods. Based on patient's CT scan data, two finite element models were created. Both models were composed of the severely resorbed posterior mandible with first premolar and second molar and missing second premolar and first molar, a two-unit prosthesis supported by two implants. The differences were in implants position and crown-to-implant ratio; lingualized implants creating lingually overcontoured prosthesis (Model CP2) and prosthetic-driven implants creating an excessive crown-to-implant ratio (Model PD2). A screw preload of 466.4 N and a buccal occlusal load of 262 N were applied. The contacts between the implant components were set to a frictional contact with a friction coefficient of 0.3. The maximum von Mises stress and strain and maximum equivalent plastic strain were analyzed and compared, as well as volumes of the materials under specified stress and strain ranges. Results. The results revealed that the highest maximum von Mises stress in each model was 1091 MPa for CP2 and 1085 MPa for PD2. In the cortical bone, CP2 showed a lower peak stress and a similar peak strain. Besides, volume calculation confirmed that CP2 presented lower volumes undergoing stress and strain. The stresses in implant components were slightly lower in value in PD2. However, CP2 exhibited a noticeably higher plastic strain. CONCLUSION. Prosthetic-driven implant placement might biomechanically be more advantageous than bone quantity-based implant placement that creates a buccal cantilever.

• Communications for Statistical Applications and Methods
• /
• 제26권4호
• /
• pp.411-430
• /
• 2019

Tumor development is driven by complex combinations of biological elements. Recent advances suggest that molecularly distinct subtypes of breast cancers may respond differently to pathway-targeted therapies. Thus, it is important to dissect pathway disturbances by integrating multiple molecular profiles, such as genetic, genomic and epigenomic data. However, missing data are often present in the -omic profiles of interest. Motivated by genomic data integration and imputation, we present a new statistical framework for pathway significance analysis. Specifically, we develop a new strategy for imputation of missing data in large-scale genomic studies, which adapts low-rank, structured matrix completion. Our iterative strategy enables us to impute missing data in complex configurations across multiple data platforms. In turn, we perform large-scale pathway analysis integrating gene expression, copy number, and methylation data. The advantages of the proposed statistical framework are demonstrated through simulations and real applications to breast cancer subtypes. We demonstrate superior power to identify pathway disturbances, compared with other imputation strategies. We also identify differential pathway activity across different breast tumor subtypes.

• 한국수자원학회논문집
• /
• 제41권3호
• /
• pp.341-351
• /
• 2008

최근 지구 온난화로 인한 이상기후의 영향으로 게릴라성 집중호우의 피해가 증가하고 있으므로 대하천뿐만 아니라 중 소하천에서도 홍수 예 경보의 중요성이 높아지고 있다. 기존의 홍수 예 경보 체계의 경우 유출량을 계산하는 전처리과정과 주 계산과정을 거치는 동안 많은 오차들이 발생하고, 누적되어 그 결과물(예측된 유출량) 속에 오차들이 내포되어 있다. 또한 유출모형의 적용에 필요한 매개변수들을 추정하기 위해서도 많은 실측자료가 필요하고, 많은 불확실성이 내재되어 있다. 본 연구에서는 기존의 홍수 예 경보 시스템의 문제점과 불확실성을 최대한 감소시키기 위해 ANFIS(Adaptive Neuro-Fuzzy Inference) 기법을 사용하였다. ANFIS는 신경회로망 기법을 사용한 data driven 모형으로 기존의 물리적 모형의 구축과정에서 필수적이었던 방대한 양의 물리적 자료를 배제하고 유역의 강우자료와 수위자료만으로 모형을 구축하고 수위 예측을 실시할 수 있다. 입력자료로는 시계열 강우자료와 수위자료를 사용하였고, 모형을 통하여 t+1, t+2, t+3 시간 후의 수위를 예측하였다. 탄천유역의 2003년부터 2005년까지의 강우사상을 이용하여 모형의 적용성과 타당성을 검토하였고, 2006년 실제 강우에 모형을 적용한 결과 실제 수위를 큰 오차 없이 모의할 수 있었다.

• 산업경영시스템학회지
• /
• 제32권4호
• /
• pp.184-191
• /
• 2009

In a technological driven environment, a depreciation estimate which is based on traditional life analysis results in a decelerated rate of capital recovery. This time pattern of technological growths models needs to be incorporated into life analysis framework especially in those industries experiencing fast technological changes. The approximation technique for calculating the variance can be applied to the six growth models that were selected by the degree of skewness and the transformation of the functions. For the Pearl growth model, the Gompertz growth model, and the Weibull growth model, the errors have zero mean and a constant variance over time. However, transformed models like the linearized Fisher-Pry model, the linearized Gompertz growth model, and the linearized Weibull growth model have increasing variance from zero to that point at which inflection occurs. It can be recommended that if the variance of error over time is increasing, then a transformation of observed data is appropriate.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 2005년도 ICCAS
• /
• pp.1343-1346
• /
• 2005

In the cold rolling mill, coil breakage that generated in rolling process makes the various types of troubles such as the degradation of productivity and the damage of equipment. Recent researches were done by the mechanical analysis such as the analysis of roll chattering or strip inclining and the prevention of breakage that detects the crack of coil. But they could cover some kind of breakages. The prediction of Coil breakage was very complicated and occurred rarely. We propose to build effective prediction modes for coil breakage in rolling process, based on data mining model. We proposed three prediction models for coil breakage: (1) decision tree based model, (2) regression based model and (3) neural network based model. To reduce model parameters, we selected important variables related to the occurrence of coil breakage from the attributes of coil setup by using the methods such as decision tree, variable selection and the choice of domain experts. We developed these prediction models and chose the best model among them using SEMMA process that proposed in SAS E-miner environment. We estimated model accuracy by scoring the prediction model with the posterior probability. We also have developed a software tool to analyze the data and generate the proposed prediction models either automatically and in a user-driven manner. It also has an effective visualization feature that is based on PCA (Principle Component Analysis).

• 한국방재안전학회논문집
• /
• 제12권2호
• /
• pp.73-82
• /
• 2019

최근 이상기후로 인한 집중호우에 따른 하천변 사회기반시설의 침수피해가 증가하고 있으며, 침수 가능성 여부에 대한 신속한 예 경보가 필요한 실정이다. 일반적인 홍수 예 경보는 하천수위를 이용하고 있으며, 수치모형을 이용하여 하천수위를 예측하는 연구가 대부분이었다. 그러나 수치모형을 이용한 하천수위 예측은 결과가 정확한 반면 수치모의 시간이 오래 소요된다는 한계점이 있어 최근에는 인공신경망 등을 적용한 자료기반의 수위예측 모형이 많이 이용되고 있다. 하지만 기존의 인공신경망을 활용한 수위예측 연구는 시간적 매개변수를 고려하지 못하였다는 한계점이 존재한다. 본 연구에서는 시간적 매개변수(Time delay= 2시간)를 고려한 NARX 신경망 모형을 사용하여 한강대교의 수위를 예측하였다. 또한 NARX 모형의 적합성을 판단하기 위하여 인공신경망(ANN) 모형과, 순환신경망(RNN)모형의 결과와 비교하였다. 2009년에서 2018년까지 10년간의 수문자료를 이용하여 70%를 학습시키고 검정과 평가에 15%를 사용하여 2018년의 한강대교 3시간 후 수위를 예측한 결과 평균제곱근오차(RMSE)의 경우 ANN, RNN, NARX model이 각각 0.20 m, 0.11 m, 0.09 m, 평균절대오차(MAE)의 경우, 각각 0.12 m, 0.06 m, 0.05 m, 첨두수위 오차(Peak Error)는 각각 1.56 m, 0.55 m, 0.10 m로 나타났다. 연구 대상지역에 대한 시간적 매개변수를 고려한 예측 결과의 오차분석을 통하여 NARX 신경망 모형을 사용하는 것이 수위예측 모형 구축이 가장 적합한 것으로 나타났다. 이는 NARX 신경망 모형이 과거의 입력자료를 고려함으로써 시계열 자료의 변동 추세도 학습 할 수 있으며, 또한 모형 내 활성함수를 쌍곡선탄젠트(Hyperbolic tangent) 및 Rectified Linear Unit(ReLU) 함수를 사용하여 고수위 예측 시에도 정확한 예측 값을 도출할 수 있기 때문이다. 그러나 NARX 신경망 모형은 시퀀스 길이가 길어짐에 따라 기울기 소실문제(Vanishing gradient)가 발생하는 한계점이 있어 향후에는 이를 보완한 LSTM(Long Short Term Model)모형을 이용하여 수위예측의 정확도를 검토하고자 한다.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제17권12호
• /
• pp.3330-3344
• /
• 2023

This paper proposes an antenna performance prediction model in the autonomous driving radar manufacturing process. Our research work is based upon a challenge dataset, Driving Radar Manufacturing Process Dataset, and a typical AutoML machine learning workflow engine, Pycaret open-source Python library. Note that the dataset contains the total 70 data-items, out of which 54 used as input features and 16 used as output features, and the dataset is properly built into resolving the multi-output regression problem. During the data regression analysis and preprocessing phase, we identified several input features having similar correlations and so detached some of those input features, which may become a serious cause of the multicollinearity problem that affect the overall model performance. In the training phase, we train each of output-feature regression models by using the AutoML approach. Next, we selected the top 5 models showing the higher performances in the AutoML result reports and applied the ensemble method so as for the selected models' performances to be improved. In performing the experimental performance evaluation of the regression prediction model, we particularly used two metrics, MAE and RMSE, and the results of which were 0.6928 and 1.2065, respectively. Additionally, we carried out a series of experiments to verify the proposed model's performance by comparing with other existing models' performances. In conclusion, we enhance accuracy for safer autonomous vehicles, reduces manufacturing costs through AutoML-Pycaret and machine learning ensembled model, and prevents the production of faulty radar systems, conserving resources. Ultimately, the proposed model holds significant promise not only for antenna performance but also for improving manufacturing quality and advancing radar systems in autonomous vehicles.

• Journal of Information Processing Systems
• /
• 제10권3호
• /
• pp.412-428
• /
• 2014

Request oriented sensor networks have stricter requirements than conventional event-driven or periodic report models. Therefore, in this paper we propose a minimum energy data aggregation (MEDA), which meets the requirements for request oriented sensor networks by exploiting a low power real-time scheduler, on-demand time synchronization, variable response frame structure, and adaptive retransmission. In addition we introduce a test bed consisting of a number of MEDA prototypes, which support near real-time bidirectional sensor networks. The experimental results also demonstrate that the MEDA guarantees deterministic aggregation time, enables minimum energy operation, and provides a reliable data aggregation service.

• 대한기계학회논문집
• /
• 제19권3호
• /
• pp.741-750
• /
• 1995

The turbulent buoyancy-driven flow in 2-dimensional enclosed cavities heated from the vertical side is numerically calculated for both cases of a Rayleigh number of 5*10$^{10}$ for air and 2.5*10$^{10}$ for water. Three different turbulence models are considered : standard k-.epsilon. model of Ozoe and low-Reynolds-number model of Lam and Bremhorst, and another low-Reynolds-number model of Davidson. The results indicate that the use of low-Reynolds number models is recommended for the indoor airflow computation, and the results from Davidson model are reasonably close to the reported experimental data. A sensitivity study shows that the amounts of wall-heat transfer and the velocity profiles with the Lam and Bremhorst model largely depend on the choice of the wall function for .epsilon..