• Proceedings of the Korea Water Resources Association Conference
• /
• 2023.05a
• /
• pp.358-358
• /
• 2023

본 연구에서는 그동안 수자원분야 강우유출 해석분야에 활용되었던 대표적인 머신러닝&딥러닝(ML&DL) 모델을 활용하여 모델의 하이퍼파라미터 튜닝뿐만 아니라 모델의 특성을 고려한 기상 및 수문데이터의 조합과 전처리(lag-time, 이동평균 등)를 통하여 데이터 특성과 ML&DL모델의 조합시나리오에 따른 일 유입량 예측성능을 비교 검토하는 연구를 수행하였다. 이를 위해 소양강댐 유역을 대상으로 1974년에서 2021년까지 축적된 기상 및 수문데이터를 활용하여 1) 강우, 2) 유입량, 3) 기상자료를 주요 영향변수(독립변수)로 고려하고, 이에 a) 지체시간(lag-time), b) 이동평균, c) 유입량의 성분분리조건을 적용하여 총 36가지 시나리오 조합을 ML&DL의 입력자료로 활용하였다. ML&DL 모델은 1) Linear Regression(LR), 2) Lasso, 3) Ridge, 4) SVR(Support Vector Regression), 5) Random Forest(RF), 6) LGBM(Light Gradient Boosting Model), 7) XGBoost의 7가지 ML방법과 8) LSTM(Long Short-Term Memory models), 9) TCN(Temporal Convolutional Network), 10) LSTM-TCN의 3가지 DL 방법, 총 10가지 ML&DL모델을 비교 검토하여 일유입량 예측을 위한 가장 적합한 데이터 조합 특성과 ML&DL모델을 성능평가와 함께 제시하였다. 학습된 모형의 유입량 예측 결과를 비교·분석한 결과, 소양강댐 유역에서는 딥러닝 중에서는 TCN모형이 가장 우수한 성능을 보였고(TCN>TCN-LSTM>LSTM), 트리기반 머신러닝중에서는 Random Forest와 LGBM이 우수한 성능을 보였으며(RF, LGBM>XGB), SVR도 LGBM수준의 우수한 성능을 나타내었다. LR, Lasso, Ridge 세가지 Regression모형은 상대적으로 낮은 성능을 보였다. 또한 소양강댐 댐유입량 예측에 대하여 강우, 유입량, 기상계열을 36가지로 조합한 결과, 입력자료에 lag-time이 적용된 강우계열의 조합 분석에서 세가지 Regression모델을 제외한 모든 모형에서 NSE(Nash-Sutcliffe Efficiency) 0.8이상(최대 0.867)의 성능을 보였으며, lag-time이 적용된 강우와 유입량계열을 조합했을 경우 NSE 0.85이상(최대 0.901)의 더 우수한 성능을 보였다.

• Journal of Korea Water Resources Association
• /
• v.57 no.7
• /
• pp.437-449
• /
• 2024

Rainfall-runoff prediction studies using deep learning while considering catchment attributes have been gaining attention. In this study, we selected two models: the Transformer model, which is suitable for large-scale data training through the self-attention mechanism, and the LSTM-based multi-state-vector sequence-to-sequence (LSTM-MSV-S2S) model with an encoder-decoder structure. These models were constructed to incorporate catchment attributes and predict the inflow of 10 multi-purpose dam watersheds in South Korea. The experimental design consisted of three training methods: Single-basin Training (ST), Pretraining (PT), and Pretraining-Finetuning (PT-FT). The input data for the models included 10 selected watershed attributes along with meteorological data. The inflow prediction performance was compared based on the training methods. The results showed that the Transformer model outperformed the LSTM-MSV-S2S model when using the PT and PT-FT methods, with the PT-FT method yielding the highest performance. The LSTM-MSV-S2S model showed better performance than the Transformer when using the ST method; however, it showed lower performance when using the PT and PT-FT methods. Additionally, the embedding layer activation vectors and raw catchment attributes were used to cluster watersheds and analyze whether the models learned the similarities between them. The Transformer model demonstrated improved performance among watersheds with similar activation vectors, proving that utilizing information from other pre-trained watersheds enhances the prediction performance. This study compared the suitable models and training methods for each multi-purpose dam and highlighted the necessity of constructing deep learning models using PT and PT-FT methods for domestic watersheds. Furthermore, the results confirmed that the Transformer model outperforms the LSTM-MSV-S2S model when applying PT and PT-FT methods.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.64 no.3
• /
• pp.393-398
• /
• 2015

City or Community-scale Energy Management System(CEMS) is used to reduce the total energy consumed in the city by arranging the energy resources efficiently at the planning stage and controlling them economically at the operating stage. Of the operational functions of the CEMS, generation forecasting of renewable energy resources is an essential feature for the effective supply scheduling. This is because it can develop daily operating schedules of controllable generators in the city (e.g. diesel turbine, micro-gas turbine, ESS, CHP and so on) in order to minimize the inflow of the external power supply system, considering the amount of power generated by the uncontrollable renewable energy resources. This paper is written to introduce numerical models for photo-voltaic power generation prediction based on the weather forecasting information. Unlike the conventional methods using the average radiation or average utilization rate, the proposed models are developed for CEMS applications using the realtime weather forecast information provided by the National Weather Service.

• Journal of Korean Society on Water Environment
• /
• v.28 no.3
• /
• pp.359-366
• /
• 2012

Accurate predictions about the water quality of a river have great importance in identifying in-stream flow and water supply requirements and solving relevant environmental problems. In this study, the effect of water release from upstream dam on the downstream water quality has been investigated by applying a hydological model combined with QUAL2E to Geum River basin. The ESP (Ensemble Stream Prediction) method, which has been validated and verified by lots of researchers, was used to predict reservoir and tributary inflow. The input parameters for a combined model to predict both hydrological characteristics and water quality were identified and optimized. In order to verify the model performance, the simulated result at Gongju station, located at the downstream from Daecheong Dam, has been compared with measured data in 2008. As a result, it was found that the proposed model simulates well the values of BOD, T-N, and T-P with an acceptable reliability.

• Journal of Korean Society of Water and Wastewater
• /
• v.9 no.4
• /
• pp.73-86
• /
• 1995

The propriety of the numerical model application was examined on Paldang resevoir and its inflow tributaries located in the center of the Korean peninsula and the long term water quality forecast of the oxygen profile was carried out in this syduy. The input data of the model was the capacity of the reservoir, catchment area, percolation, diffusion rate, vertical mixing rate, dissolution rate from the bottom of the reservoir, outflow of the resevoir, water quality measurement and meteorology data of the drainage basin, and the output result was the annual estimation value of the dissolved oxygen concentration and the biochemical oxygen demand. The modeling method is based on the measured or calculated boundary condition dividing the water area into several blocks from the macorscopic aspect and considering the mass balance in these blocks. As the result of the water quality forecast, it was expected that the water quality in Northern Han River and Paldang reservoir would maintain the recent level, but that the water quality in the Southern Han River and its inflow tributary would worsen below the grade 4 of the life environmental standard from around 2000 owing to the decrease of DO concentration and the increase of BOD concentration.

• Journal of the Korean GEO-environmental Society
• /
• v.3 no.4
• /
• pp.11-17
• /
• 2002

The purpose of this study is the prediction of pollutant loads into Seomjin river watershed using neural networks model. The pollutant loads into river watershed depend upon the water quantity of inflow from the upstream as well as the water quality of the inflow into the river. For the estimation of pollutants into river, a neural networks model which has the features of multi-layered structure and parallel multi-connections is used. The used water quality parameters are BOD, COD and SS into Seomjin river. The results of calibration are satisfactory, and proved the availability of a proposed neural networks model to estimate short-term water quality pollutants into river system.

• Publications of The Korean Astronomical Society
• /
• v.30 no.2
• /
• pp.93-97
• /
• 2015

How high-mass stars form is currently unclear. Calculations suggest that the radiation pressure of a forming star can halt spherical infall, preventing further growth when it reaches $10M_{\odot}$. Two major theoretical models on the further growth of stellar mass have been proposed. One model suggests the merging of less massive stellar objects, and the other is through accretion, but with the help of a disk. Inflow motions are key evidence for how forming stars gain further mass to build up massive stars. Recent developments in technology have boosted the search for inflow motion. A number of high-mass collapse candidates were obtained with single dish observations, and mostly showed blue profiles. Infalling signatures seem to be more common in regions which have developed radiation pressure than in younger cores, which is the opposite of the theoretical prediction and is also very different from observations of low mass star formation. Interferometer studies so far confirm this tendency with more obvious blue profiles or inverse P Cygni profiles. Results seem to favor the accretion model. However, the evolution of the infall motion in massive star forming cores needs to be further explored. Direct evidence for monolithic or competitive collapse processes is still lacking. ALMA will enable us to probe more detail of the gravitional processes.

• Journal of Environmental Science International
• /
• v.26 no.9
• /
• pp.1031-1043
• /
• 2017

Recently severe drought caused the water shortage around the western parts of Chungcheongnamdo province, South Korea. A Diversion tunnel from the Geum river to the Boryong dam, which is the water supply dam for these areas has been proposed to solve this problem. This study examined hydraulic impacts on the Geum river associated with the diversion plan assuming the severe drought condition of 2015 would persist for the simulation period of 2016. The hydraulic simulation model was verified using hydrologic and hydraulic data including hourly discharges of the Geum river and its 8 tributaries, fluctuation of tidal level at the mouth of the river, withdrawals and return flows and operation records of the Geum river barrage since Feb. 1, 2015 through May 31, 2015. For the upstream boundary condition of the Geum river predicted inflow series using the nonlinear regression equation for 2015 discharge data was used. In order to estimate the effects of uncertainty in inflow prediction to the results total four inflow series consisting of upper limit flow, expected flow, lower limit flow and instream flow were used to examine hydraulic impacts of the diversion plan. The simulation showed that in cases of upper limit and expected flows there would be no problem in taking water from the Geum river mouth with a minimum water surface level of EL(+) 1.44 m. Meanwhile, the simulation also showed that in cases of lower limit flow and instream flow there would be some problems not only in taking water for water supply from the mouth of the Geum river but also operating the diversion facility itself with minimum water surface levels of EL(+) 0.94, 0.72, 0.43, and 0.14 m for the lower limit flow without/with diversion and the instream flow without/with diversion, respectively.

• Magazine of the Korean Society of Agricultural Engineers
• /
• v.35 no.1
• /
• pp.50-58
• /
• 1993

The sedimentation patterns at a reservoir, important to the reservoir capacity curve were simulated using a depth averaged, two-dimensional sediment transport model, that is capable of depicting velocity distributions and sediment transportation. The Banweol reservoir, whose stage capacity relationships have been surveyed before and after the construction, was selected and the daily inflow rates and stages were simulated using a reservoir operation model(DI-ROM). The applicability of the transport model was tested from the comparisons of simulated sedimentation patterns to the surveyed results. The simulated inflow rates and water level fluctuations at the reservoir during twenty-one years from 1966 to 1986, showed that water levels exceeding 80 percent of the total capacity occurred for 70 percent of the periods and inflow rates less than 5000rn$^3$/day sustained for 54 percent of the spans. Dorminant flow directions were simulated from two streamflow inlets to the dam site. And simulated sediment concentrations were higher near the inlets and lower at the inside of the reservoir. Sediment was deposited heavily near the inlets, and portions of sediments were distributed along the flow paths within the reservoir. The comparisons between the simulation results and the surveyed depositions were partially matched. However, it was not possible to compare two results at the upper parts of the reservoir where dredging was carried out few times for the purpose of reservoir maintenance. This study demonstrates that sedimentation patterns within the reservoir are closely related to incoming sediment and flow rates, water level fluctuations, and flow circulation within the reservoir.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2022.05a
• /
• pp.307-307
• /
• 2022

최근 기후변화와 댐 상류 토지이용 변화 등과 같은 다양한 원인에 의해 댐 유입량의 변동성이 증가하면서 댐 관리 및 운영조작 의사 결정에 어려움이 발생하고 있다. 따라서 이러한 댐 유입량의 변동 특성을 반영하여 댐 유입량을 정확하고 효율적으로 예측할 수 있는 방안이 필요한 실정이다. 머신러닝 기술이 발전하면서 Auto-ML(Automated Machine Learning)이 다양한 분야에서 활용되고 있다. Auto-ML은 데이터 전처리, 최적 알고리즘 선택, 하이퍼파라미터 튜닝, 모델 학습 및 평가 등의 모든 과정을 자동화하는 기술이다. 그러나 아직까지 수문 분야에서 댐 유입량을 예측하기 위한 모델을 개발하는데 있어서 Auto-ML을 활용한 사례는 부족하고, 특히 댐 유입량의 예측 정확성을 확보하기 위해 High-inflow and low-inflow 의 변동 특성을 고려한 하이브리드 결합 방식을 통해 Auto-ML 기반 앙상블 모델을 개발하고 평가한 연구는 없다. 본 연구에서는 Auto-ML의 패키지 중 Auto-sklearn을 통해 홍수기, 비홍수기 유입량 변동 특성을 반영한 하이브리드 앙상블 댐 유입량 예측 모델을 개발하였다. 소양강댐을 대상으로 적용한 결과, 하이브리드 Auto-sklearn 앙상블 모델의 댐 유입량 예측 성능은 R² 0.868, RMSE 66.23 m³/s, MAE 16.45 m³/s로 단일 Auto-sklearn을 통해 구축 된 앙상블 모델보다 전반적으로 우수한 것으로 나타났다. 특히 FDC (Flow Duration Curve)의 저수기, 갈수기 구간에서 두 모델의 유입량 예측 경향은 큰 차이를 보였으며, 하이브리드 Auto-sklearn 모델의 예측 값이 관측 값과 더욱 유사한 것으로 나타났다. 이는 홍수기, 비홍수기 구간에 대한 앙상블 모델이 독립적으로 구축되는 과정에서 각 모델에 대한 하이퍼파라미터가 최적화되었기 때문이라 판단된다. 향후 본 연구의 방법론은 보다 정확한 댐 유입량 예측 자료를 생성하기 위한 방안 수립뿐만 아니라 다양한 분야의 불균형한 데이터셋을 이용한 앙상블 모델을 구축하는데도 유용하게 활용될 수 있을 것으로 사료된다.