Various elements of Fabrication (FAB), mass production of existing products, new product development and process improvement evaluation might increase the complexity of production process when products are produced at the same time. As a result, complex production operation makes it difficult to predict production capacity of facilities. In this environment, production forecasting is the basic information used for production plan, preventive maintenance, yield management, and new product development. In this paper, we tried to develop a multiple linear regression analysis model in order to improve the existing production capacity forecasting method, which is to estimate production capacity by using a simple trend analysis during short time periods. Specifically, we defined overall equipment effectiveness of facility as a performance measure to represent production capacity. Then, we considered the production capacities of interrelated facilities in the FAB production process during past several weeks as independent regression variables in order to reflect the impact of facility maintenance cycles and production sequences. By applying variable selection methods and selecting only some significant variables, we developed a multiple linear regression forecasting model. Through a numerical experiment, we showed the superiority of the proposed method by obtaining the mean residual error of 3.98%, and improving the previous one by 7.9%.

고준위방사성폐기물의 처분은 고심도 암반내에 처분시스템을 구축하는 심층 처분방법이 고려된다. 심층 처분은 처분용기, 완충재, 뒷채움재, 근계암반의 설계 요소인 공학적방벽과 천연 방벽으로 구성된다. 공학적방벽 중에서 벤토나이트 완충재는 암반으로부터 유입되는 지하수 흐름을 최소화하고 핵종 유출을 저지하는 기능을 한다. 지하수 유입으로 인한 완충재의 수리전도도 특성 규명은 처분장 공학적방벽의 안정성 및 건전성에 대한 성능 평가에 있어 중요한 사안이다. 본 연구에서는 경주 벤토나이트를 이용하여 다양한 건조밀도와 온도 조건에 따라 포화 수리전도도 실험을 수행하였으며, 120개의 실험 결과 를 다중 회귀 분석을 통해 수리전도도 추정 모델을 제시하였다. 실험 결과에서는 건조밀도가 커질수록 수리전도도가 감소하는 경향이 나타났다. 또한, 온도가 증가할수록 수리전도도가 증가하였다. 이러한 실험 결과들을 종합한 다중 회귀 분석 결과에서는 수리전도도 추정식의 결정계수(R2)가 0.93으로 높게 나타났다. 본 연구에서 제시된 수리전도도 추정식은 벤토나이트 완충재의 성능과 연관된 건조밀도와 온도의 영향을 고려하여 처분시스템의 공학적방벽 설계에 활용 될 것으로 판단된다.

Supply chain managers seek to achieve global optimization by solving problems in the supply chain's business process. However, companies in the supply chain hide the adverse information and inform only the beneficial information, so the information is distorted and cannot be the information that describes the entire supply chain. In this case, supply chain managers can directly collect and analyze supply chain activity data to find and manage the companies described by the data. Therefore, this study proposes a method to collect the order-inventory information from each company in the supply chain and detect the companies whose data characteristics are explained through deep learning. The supply chain consists of Manufacturer, Distributor, Wholesaler, Retailer, and training and testing data uses 600 weeks of time series inventory information. The purpose of the experiment is to improve the detection accuracy by adjusting the parameter values of the deep learning network, and the parameters for comparison are set by learning rate (lr = 0.001, 0.01, 0.1) and batch size (bs = 1, 5). Experimental results show that the detection accuracy is improved by adjusting the values of the parameters, but the values of the parameters depend on data and model characteristics.

In this paper, the model for predicting yields of chinese cabbages of each cultivar (joined-up in 2015 and wrapped-up in 2016) was developed after the reflectance of hyperspectral imagery was merged as 10 nm, 25 nm and 50 nm of FWHM (full width at half maximum). Band rationing was employed to minimize the unstable reflectance of multi-temporal hyperspectral imagery. The stepwise analysis was employed to select key band ratios to predict yields in all cultivars. The key band ratios selected for each of FWHM were used to develop the yield prediction models of chinese cabbage for all cultivars (joined-up & wrapped-up) and each cultivar (joined-up, wrapped-up). Effective accumulated temperature (EAT) was added in the models to evaluate its improvement of performances. In all models, the performance of models was improved with adding of EAT. The models with EAT for each of FWHM showed the predictability of yields in all cultivars as R2≥0.80, RMSE≤694 g/plant and RE≤28.3%. Such as this result, if the yield can be predicted regardless of the cultivar, it is considered to be advantageous when predicting the yield over a wide area because it is not require a cultivar classification work as pre-processing in imagery.

본 연구에서는 인공지능기법을 이용하여 진동만의 용존산소량 예측을 하였다. 관측자료에 존재하는 결측 구간을 보간하기 위해 양방향재귀신경망(BRITS, Bidirectional Recurrent Imputation for Time Series) 딥러닝 알고리즘을 이용하였고, 대표적 시계열 예측 선형모델인 ARIMA(Auto-Regressive Integrated Moving Average)과 비선형모델 중 가장 많이 이용되고 있는 LSTM(Long Short-Term Memory) 모델을 이용 하여 진동만의 용존산소량을 예측하고 그 성능을 평가했다. 결측 구간 보정 실험은 표층에서 높은 정확도로 보정이 가능했으나, 저층에서는 그 정확도가 낮았으며, 중층에서는 실험조건에 따라 정확도가 불안정하게 나타났다. 실험조건에 따라 정확도가 불안정하게 나타났다. 결과로부터 LSTM 모델이 중층과 저층에서 ARIMA 모델보다 우세한 정확도를 보였으나, 표층에서는 ARIMA모델의 정확도가 약간 높은 것으로 나타났다.

Recently, deep learning that is the most popular and effective class of machine learning algorithms is widely applied to various industrial areas. A number of research on various topics about structural engineering was performed by using artificial neural networks, such as structural design optimization, vibration control and system identification etc. When nonlinear semi-active structural control devices are applied to building structure, a lot of computational effort is required to predict dynamic structural responses of finite element method (FEM) model for development of control algorithm. To solve this problem, an artificial neural network model was developed in this study. Among various deep learning algorithms, a recurrent neural network (RNN) was used to make the time history response prediction model. An RNN can retain state from one iteration to the next by using its own output as input for the next step. An eleven-story building structure with semi-active tuned mass damper (TMD) was used as an example structure. The semi-active TMD was composed of magnetorheological damper. Five historical earthquakes and five artificial ground motions were used as ground excitations for training of an RNN model. Another artificial ground motion that was not used for training was used for verification of the developed RNN model. Parametric studies on various hyper-parameters including number of hidden layers, sequence length, number of LSTM cells, etc. After appropriate training iteration of the RNN model with proper hyper-parameters, the RNN model for prediction of seismic responses of the building structure with semi-active TMD was developed. The developed RNN model can effectively provide very accurate seismic responses compared to the FEM model.

Background: Scapular winging (SW) could be caused by tightness or weakness of the periscapular muscles. Although data mining techniques are useful in classifying or predicting risk of musculoskeletal disorder, predictive models for risk of musculoskeletal disorder using the results of clinical test or quantitative data are scarce. Objects: This study aimed to (1) investigate the difference between young women with and without SW, (2) establish a predictive model for presence of SW, and (3) determine the cutoff value of each variable for predicting the risk of SW using the decision tree method. Methods: Fifty young female subjects participated in this study. To classify the presence of SW as the outcome variable, scapular protractor strength, elbow flexor strength, shoulder internal rotation, and whether the scapula is in the dominant or nondominant side were determined. Results: The classification tree selected scapular protractor strength, shoulder internal rotation range of motion, and whether the scapula is in the dominant or nondominant side as predictor variables. The classification tree model correctly classified 78.79% (p = 0.02) of the training data set. The accuracy obtained by the classification tree on the test data set was 82.35% (p = 0.04). Conclusion: The classification tree showed acceptable accuracy (82.35%) and high specificity (95.65%) but low sensitivity (54.55%). Based on the predictive model in this study, we suggested that 20% of body weight in scapular protractor strength is a meaningful cutoff value for presence of SW.

본 연구에서는 복잡한 도심의 구조로 인한 미세먼지 농도의 강화 가능성에 대하여 데이터 마이닝 기술과 군집분석을 이용해 조사하였다. 데이터 마이닝 분석에서 미세먼지 농도와 서울지역 도시용도 데이터 사이에는 유의한 상관관계를 보이지 않았다. 그러나 전국 공공데이터를 기반으로 한 군집분석에서는 건물의 높이(층수)에서 특히 PM10과 강한 상관관계가 나타났다. 단일 케노피 모델(Single Canopy Model) 및 미기상 도시모델링 프로그램(ENVI-Met.4)을 사용한 모델링 분석을 실시하여 도시지역에서 모사된 대기 대류가 건물 분포 및 높이 유형의 배열에 따라 다양한 난류의 패턴을 구현함을 확인하였 다. 도시 건물의 복잡한 구조는 대류활동을 제어하여 정체상태를 유도하고 지표 부근의 미세먼지 강화가능성을 초래 하였다. 따라서 도심 구조와 형태에 따른 열환경의 변화로 인한 정체 효과는 미세먼지 산정에 있어서 반드시 고려되어야 한다. 복잡한 도시지역의 미세먼지 잔류확률에 대한 정보를 제공하기 위해서는 대기정체 현상이 중요한 의미로 해석될 수 있다.

The influx of marine exotic and alien species is disrupting marine ecosystems and aquaculture. Herdmania momus, reported as an invasive species, is distributed all along the coast of Jeju Island and has been confirmed to be distributed and spread to Busan. The potential habitats and distribution of H. momus were estimated using the maximum entropy (MaxEnt) model, quantum geographic information system (QGIS), and Bio-ocean rasters for analysis of climate and environment (Bio-ORACLE), which can predict the distribution and spread based only on species occurrence data using species distribution model (SDM). Temperature and salinity were selected as environmental variables based on previous literature. Additionally, two different representative concentration pathway (RCP) scenarios (RCP 4.5 and RCP 8.5) were set up to estimate future and potential habitats owing to climate change. The prediction of potential habitats and distribution for H. momus using MaxEnt confirmed maximum temperature as the highest contributor (77.1%), and mean salinity, the lowest (0%). And the potential habitats and distribution of H. momus were the highest on Jeju Island, and no potential habitat or distribution was seen in the Yellow Sea. Different RCP scenarios showed that at RCP 4.5, H. momus would be distributed along the coast of Jeju Island in the year 2050 and that the distribution would expand to parts of the Korea Strait by the year 2100. RCP 8.5, the distribution in 2050 is predicted to be similar to that at RCP 4.5; however, by 2100, the distribution is predicted to expand to parts of the Korea Strait and the East Sea. This study can be utilized as basic data to effectively control the ecological injuries by H. momus by predicting its spread and distribution both at present and in the future.

There have been a lot of studies in the past for the method of predicting the failure of a machine, and recently, a lot of researches and applications have been generated to diagnose the physical condition of the machine and the parts and to calculate the remaining life through various methods. Survival models are also used to predict plant failures based on past anomaly cycles. In particular, special machine that reflect the fluid flow and process characteristics of chemical plants are connected to hundreds or thousands of sensors, so there are not many factors that need to be considered, such as process and material data as well as application of derivative variables. In this paper, the data were preprocessed through time series anomaly detection based on unsupervised learning to predict the abnormalities of these special machine. Next, clustering results reflecting clustering-based data characteristics were applied to produce additional variables, and a learning data set was created based on the history of past facility abnormalities. Finally, the prediction methodology based on the supervised learning algorithm was applied, and the model update was confirmed to improve the accuracy of the prediction of facility failure. Through this, it is expected to improve the efficiency of facility operation by flexibly replacing the maintenance time and parts supply and demand by predicting abnormalities of machine and extracting key factors.

The objectives of this study were to estimate genetic parameters and breeding values of four carcass traits of the Hanwoo cattle breed: carcass weight (CWT), back fat thickness (BFT), eye-muscle area (EMA), and marbling score (MAR). Genetic parameters and breeding values were estimated based on data (“estimating dataset”) collected from September 2004 to March 2019. Predictability of parental breeding value estimates (EBVs) for the performances of progeny of the control group was evaluated on another dataset (“testing dataset”) using linear model equations involving parental EBVs classified into sex and age groups. The parental EBVs of animals in the testing dataset were traced by pedigree relationships of animals in the estimating dataset. Heritability estimates of CWT, BFT, EMA, and MAR were 0.53, 0.43, 0.38, and 0.54, respectively. Genetic correlation coefficients of CWT with BFT, EMA, and MAR were +0.32, +0.59, and +0.11, respectively. Environmental correlation coefficients of CWT with BFT, EMA, and MAR were +0.46, +0.55, and +0.29, respectively. In the testing dataset, partial regression coefficients of phenotypic values of progeny on sire EBVs ranged from +0.43 to +0.60 depending on traits fit into the models, while those on dam EBVs ranged from +0.54 to +0.67. All partial regression coefficients were statistically significant and were approximated to the expected value of +0.5. Together, these values validate the use of parental EBVs for predicting progeny carcass phenotypes in the Hanwoo herd.

본 연구에서는 당동만을 중심으로 빈산소가 발생하는 물리적 해양환경 특성을 파악하고, 로지스틱 회귀분석을 이용해 빈 산소 발생확률을 예측하였다. 관측 자료를 분석한 결과, 브런트-바이살라 주파수는 수심이 깊은 만 입구보다 수심이 얕은 만 내측에서 더 크게 나타났다. 이는 당동만 내측에서 담수 유입으로 인해 표층 염분이 낮아져 강한 밀도 성층이 형성되었기 때문이다. 시간적으로 는 6월 ~ 9월까지 리차드슨 수와 브런트 바이살라 주파수가 매우 높게 나타났고, 9월 2일 이후로는 성층이 완화되어 감소하는 경향을 보였다. 당동만에서 관측된 용존산소 및 수온, 염분 자료를 분석한 결과, 저층의 용존산소 농도는 공통적으로 표층과 저층의 수온차에 가장 큰 영향을 받는 것으로 나타났다. 한편, 수심차(dz)를 고정된 변수로 두고, 수온차(dt)의 변화에 의한 빈산소의 발생 확률의 변화 를 계산한 결과, 수심차(dz)가 각각 5 m, 10 m, 15 m, 20 m일 경우, 수온차(dt)는 8℃, 7℃, 5℃, 3℃일 때 빈산소 발생확률이 70 %를 상회 하는 것으로 나타났다. 이는 당동만에서 수심차(dz)가 커질수록 빈산소 발생에 필요한 수온차(dt)는 작아지게 된다는 것을 뜻하며, 특 히 당동만에서 수심차(dz)가 20 m 내외인 지역은 빈산소가 발생하기 매우 쉬운 환경이라는 것을 알 수 있었다.

Turbidity has various effects on the water quality and ecosystem of a river. High turbidity during floods increases the operation cost of a drinking water supply system. Thus, the management of turbidity is essential for providing safe water to the public. There have been various efforts to estimate turbidity in river systems for proper management and early warning of high turbidity in the water supply process. Advanced data analysis technology using machine learning has been increasingly used in water quality management processes. Artificial neural networks(ANNs) is one of the first algorithms applied, where the overfitting of a model to observed data and vanishing gradient in the backpropagation process limit the wide application of ANNs in practice. In recent years, deep learning, which overcomes the limitations of ANNs, has been applied in water quality management. LSTM(Long-Short Term Memory) is one of novel deep learning algorithms that is widely used in the analysis of time series data. In this study, LSTM is used for the prediction of high turbidity(>30 NTU) in a river from the relationship of turbidity to discharge, which enables early warning of high turbidity in a drinking water supply system. The model showed 0.98, 0.99, 0.98 and 0.99 for precision, recall, F1-score and accuracy respectively, for the prediction of high turbidity in a river with 2 hour frequency data. The sensitivity of the model to the observation intervals of data is also compared with time periods of 2 hour, 8 hour, 1 day and 2 days. The model shows higher precision with shorter observation intervals, which underscores the importance of collecting high frequency data for better management of water resources in the future.

We apply a modified Convolutional Neural Network (CNN) model in conjunction with transfer learning to predict whether an active region (AR) would produce a ≥C-class or ≥M-class flare within the next 24 hours. We collect line-of-sight magnetogram samples of ARs provided by the SHARP from May 2010 to September 2018, which is a new data product from the HMI onboard the SDO. Based on these AR samples, we adopt the approach of shuffle-and-split cross-validation (CV) to build a database that includes 10 separate data sets. Each of the 10 data sets is segregated by NOAA AR number into a training and a testing data set. After training, validating, and testing our model, we compare the results with previous studies using predictive performance metrics, with a focus on the true skill statistic (TSS). The main results from this study are summarized as follows. First, to the best of our knowledge, this is the first time that the CNN model with transfer learning is used in solar physics to make binary class predictions for both ≥C-class and ≥M-class flares, without manually engineered features extracted from the observational data. Second, our model achieves relatively high scores of TSS = 0.640±0.075 and TSS = 0.526±0.052 for ≥M-class prediction and ≥C-class prediction, respectively, which is comparable to that of previous models. Third, our model also obtains quite good scores in five other metrics for both ≥C-class and ≥M-class flare prediction. Our results demonstrate that our modified CNN model with transfer learning is an effective method for flare forecasting with reasonable prediction performance.