• 한국정보통신학회:학술대회논문집
• /
• 한국정보통신학회 2022년도 춘계학술대회
• /
• pp.491-493
• /
• 2022

세계 각지에서 집중호우, 태풍 등으로 인한 홍수 피해가 많이 발생하고 있으며, 이러한 피해를 줄이기 위해 홍수를 미리 예측하는 것은 수해 피해 관리 차원에서 필수적인 요소이다. 본 논문에서는 홍수예측을 위한 핵심 파라미터인 수위, 강수량, 그리고 습도 데이터를 입력 데이터로 활용한 수위 예측 모델을 제안한다. 많은 연구 분야에서 이미 시계열 데이터 예측 성능이 검증된 LSTM 및 GRU 모델을 기반으로 기상청에서 제공하는 종관기상관측 자료와, 방재기상관측 자료를 활용하여 입력 데이터셋을 다르게 구축하고, 성능 비교 실험을 진행하였다. 결과적으로 종관기상관측 자료를 사용했을 때 가장 좋은 결과를 얻었다. 본 논문을 통해 입력 데이터에 따른 성능 비교 실험을 진행하였고, 향후 연구로 홍수 위험도 판별 모델과 연계하여 사전에 대피 결정이 가능한 시스템 개발의 초기 연구로서 활용될 수 있을 것으로 사료된다.

• 한국농공학회논문집
• /
• 제62권6호
• /
• pp.1-9
• /
• 2020

Reservoir storage and water level information is essential for accurate drought monitoring and prediction. In particular, the agricultural drought has increased the risk of agricultural water shortages due to regional bias in reservoirs and water supply facilities, which are major water supply facilities for agricultural water. Therefore, it is important to evaluate the available water capacity of the reservoir, and it is necessary to determine the water surface area and water capacity. Remote sensing provides images of temporal water storage and level variations, and a combination of both measurement techniques can indicate a change in water volume. In areas of ungauged water volume, satellite remote sensing image acts as a powerful tool to measure changes in surface water level. The purpose of this study is to estimate of reservoir storage and level variations using satellite remote sensing image combined with hydrological statistical data and the Normalized Difference Water Index (NDWI). Water surface areas were estimated using the Sentinel-2 satellite images in Seosan, Chungcheongnam-do from 2016 to 2018. The remote sensing-based reservoir storage estimation algorithm from this study is general and transferable to applications for lakes and reservoirs. The data set can be used for improving the representation of water resources management for incorporating lakes into weather forecasting models and climate models, and hydrologic processes.

• 상하수도학회지
• /
• 제19권5호
• /
• pp.572-581
• /
• 2005

This research is conducted to develop predictable method of real scale nanofiltration treatability with small scale nanofiltration experiments. As a result of comparing calculated values with measured values, they are in a good agreement for the concentrations in filtered water and concentrated water. The results of that are not affected by change of system recovery from 20% to 95%. The proposed method is produced using constant recovery of elements, that is, no considering the pressure change. we can predict filtrated flux and contaminant concentrations with the method. The method has the following steps. (1) Calculate recovery of each element with water quality level after fixing recovery elements, (2) Predict system recovery with recovery of elements in 1, 2, 3, and 4 banks, (3) Run small scale nanofiltration experiments in predicted water quality and (4) Simulate large scale nanofiltration system for forecasting actual water quality. As the cost for nanofiltration pretest will reduced if we use the proposed method, it will be a promising method for introducing nanofiltration to supply safe drinking water.

• Geomechanics and Engineering
• /
• 제38권1호
• /
• pp.69-77
• /
• 2024

The development of shield-driven cross-river tunnels in China is witnessing a notable shift towards larger diameters, longer distances, and higher water pressures due to the more complex excavation environment. Complex geological formations, such as fault and karst cavities, pose significant construction risks. Real-time adjustment of shield tunneling parameters based on parameter prediction is the key to ensuring the safety and efficiency of shield tunneling. In this study, prediction models for the torque and thrust of the cutter plate of ultra-large diameter slurry shield TBMs is established based on integrated learning algorithms, by analyzing the real data of Heyan Road cross-river tunnel. The influence of geological complexities at the excavation face, substantial burial depth, and high water level on the slurry shield tunneling parameters are considered in the models. The results reveal that the predictive models established by applying Random Forest and AdaBoost algorithms exhibit strong agreement with actual data, which indicates that the good adaptability and predictive accuracy of these two models. The models proposed in this study can be applied in the real-time prediction and adaptive adjustment of the tunneling parameters for shield tunneling under complex geological conditions.

• 한국수자원학회논문집
• /
• 제55권11호
• /
• pp.903-911
• /
• 2022

물을 공급하기 위한 자원 중 하나인 지하수는 다양한 자연적 요인에 의해 수위의 변동이 발생한다. 최근, 인공신경망을 이용하여 지하수위의 변동을 예측하는 연구가 진행되었다. 기존에는 인공신경망 연산자 중 학습에 영향을 미치는 Optimizer로 경사하강법(Gradient Descent, GD) 기반 Optimizer를 사용하였다. GD 기반 Optimizer는 초기 상관관계 의존성과 해의 비교 및 저장 구조 부재의 단점이 존재한다. 본 연구는 GD 기반 Optimizer의 단점을 개선하기 위해 GD와 화음탐색법(Harmony Search, HS)를 결합한 새로운 Optimizer인 Gradient Descent combined with Harmony Search(GDHS)를 개발하였다. GDHS의 성능을 평가하기 위해 다층퍼셉트론(Multi Layer Perceptron, MLP)을 이용하여 이천율현 관측소의 지하수위를 학습 및 예측하였다. GD 및 GDHS를 사용한 MLP의 성능을 비교하기 위해 Mean Squared Error(MSE) 및 Mean Absolute Error(MAE)를 사용하였다. 학습결과를 비교하면, GDHS는 GD보다 MSE의 최대값, 최소값, 평균값 및 표준편차가 작았다. 예측결과를 비교하면, GDHS는 GD보다 모든 평가지표에서 오차가 작은 것으로 평가되었다.

• Nuclear Engineering and Technology
• /
• 제52권7호
• /
• pp.1429-1435
• /
• 2020

Applying an accurate parametric prediction model to identify abnormal or false pressurizer water levels (PWLs) is critical to the safe operation of marine pressurized water reactors (PWRs). Recently, deep-learning-based models have proved to be a powerful feature extractor to perform high-accuracy prediction. However, the effectiveness of models still suffers from two issues in PWL prediction: the correlations shifting over time between PWL and other feature parameters, and the example imbalance between fluctuation examples (minority) and stable examples (majority). To address these problems, we propose a cost-sensitive mechanism to facilitate the model to learn the feature representation of later examples and fluctuation examples. By weighting the standard mean square error loss with a cost-sensitive factor, we develop a Cost-Sensitive Long Short-Term Memory (CSLSTM) model to predict the PWL of PWRs. The overall performance of the CSLSTM is assessed by a variety of evaluation metrics with the experimental data collected from a marine PWR simulator. The comparisons with the Long Short-Term Memory (LSTM) model and the Support Vector Regression (SVR) model demonstrate the effectiveness of the CSLSTM.

• 반도체디스플레이기술학회지
• /
• 제18권4호
• /
• pp.87-91
• /
• 2019

The diagnosis of the failure for the existing electrical facilities was based on regular preventive maintenance, but this preventive maintenance was limited in preventing a lot of cost loss and sudden system failure. To overcome these shortcomings, fault prediction and diagnostic techniques are critical to increasing system reliability by monitoring electrical installations in real time and detecting abnormal conditions in the facility early. As the performance and quality deterioration problem occurs frequently due to the increase in the number of users of the motor pump, the purpose is to build an intelligent control system that can control the motor pump to maximize the performance and to improve the quality and reliability. To this end, a vibration sensor, temperature sensor, pressure sensor, and low water level sensor are used to detect vibrations, temperatures, pressures, and low water levels that can occur in the motor pump, and to build a system that can identify and diagnose information to users in real time.

• 대한전기학회논문지:전기물성ㆍ응용부문C
• /
• 제53권7호
• /
• pp.364-369
• /
• 2004

In this Paper, a flashover prediction method using the leakage current in the contaminated EPDM distribution polymer insulator is proposed. The leakage currents on the insulator were measured simultaneously with the different salt fog application such as 25g, 50g, and 75g per liter of deionized water. Then, the measured leakage currents were enveloped and transformed as the CDFS using the Hilbert transform and the level crossing rate, respectively. The obtained CDFS having different gradients(angles) were used as a important factor for the flashover prediction of the contaminated polymer insulator. Thus, the average angle change with an identical salt fog concentration was within a range of 20 degrees, and the average angle change among the different salt fog concentrations was 5 degrees. However, it is hard to be distinguished each other because the gradient differences among the CDFS were very small. So, the new weighting value was defined and used to solve this problem. Through simulation, it Is verified that the proposed method has the capability of the flashover prediction.

• 한국환경농학회지
• /
• 제27권4호
• /
• pp.314-320
• /
• 2008

During the summer season, more than half of the annual precipitation in Korea occurs during the summer season due to the geographical location in the Asian monsoon belt. So, this causes severe soil erosion from croplands, which is directly linked to the deterioration of crop/land productivity and surface water quality. Therefore, much attention has been given to develop accurate estimation tools of soil erosion. The aim of this study is to assess the performance of using the empirical Universal Soil Loss Equation (USLE) and the physical-based model of the Water Erosion Prediction Project (WEPP) to quantify eroded amount of soil from agricultural fields. Input data files, including climate, soil, slope, and cropping management, were modified to fit into Korean conditions. Chuncheon (forest) and Jeonju (level-plain) were selected as two Korean cities with different topographic characteristics for model analysis. The results of this current study indicated that better soil erosion prediction can be achieved using the WEPP model since it has better power to illustrate a higher degree of spatial variability than USLE in topography, precipitation, soils, and crop management practices. These present findings are expected to contribute to the development of the environmental assessment program as well as the conservation of the agricultural environment in Korea.

• 한국정보기술학회논문지
• /
• 제17권9호
• /
• pp.11-17
• /
• 2019

최근 들어 동해안에서 너울성 파도에 의한 손실이 빈번히 발생하고 있다. 너울성 파도는 다양한 요인들이 결합되어 발생하기 때문에 예측이 어렵다. 본 연구에서는 머신러닝 기술에 기초하여 동해안에서 너울성 파도의 발생을 예측하는 모델을 제안하였다. 모델 개발을 위해 포항 신항의 하역중단 데이터 및 신항 부근의 기압, 풍속, 풍향, 수온 등의 기상자료를 수집하였다. 수집한 데이터로부터 너울발생에 중요한 영향을 미치는 변수들을 선별하였으며, 모델 개발을 위해 다양한 머신러닝 예측 알고리즘들을 테스트 하였다. 그 결과 조위, 수온, 기압이 너울 발생 예측을 위한 주요 변수로 확인이 되었고, Random Forest 모델이 가장 우수한 성능을 보였으며. 모델의 예측 정확도는 88.6%이다.