http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
Extreme learning machine 기법을 이용한 소양강댐 월 유입량 예측
김병식(ByungSik Kim),최승철(SeungCheol Choi),이병현(ByungHyun Lee),하헌중(HernJoong Ha) 한국데이터정보과학회 2024 한국데이터정보과학회지 Vol.35 No.3
In recent years, the frequency of flooding due to heavy rainfall has been increasing due to climate change, highlighting the growing importance of disaster prevention. Accurate prediction of dam inflow is crucial during heavy rainfall and typhoons for proper dam discharge. To simulate inflow, various approaches, including physical models and machine learning models., are employed. In this study, we utilized the Extreme Learning Machine (ELM), a machine learning technique, to simulate the inflow of Soyang River Dam watershed using precipitation data observed at the weather station in Inje (Station 211) and historical inflow data. Data were collected from January 1974 to August 2023, with training using data from January 1974 to December 2020 and validation with data from January 2021 to August 2023. Additionally, we compared the results of ELM with those of a Multilayer Perceptron (MLP) with a similar structure and conducted model validation using evaluation metrics. The proposed ELM model showed validation results for the test data, achieving an MAE of 19.98, MSE of 931.25 and R-squared value of 0.83.
ROC를 이용한 보행에 영향을 미치는 한계강우량의 정확도 평가
추경수,강동호,김병식,Choo, Kyungsu,Kang, Dongho,Kim, Byungsik 한국수자원학회 2020 한국수자원학회논문집 Vol.53 No.12
Recently, as local heavy rains occur frequently in a short period of time, economic and social impacts are increasing beyond the simple primary damage. In advanced meteorologically advanced countries, realistic and reliable impact forecasts are conducted by analyzing socio-economic impacts, not information transmission as simple weather forecasts. In this paper, the degree of flooding was derived using the Spatial Runoff Assessment Tool (S-RAT) and FLO-2D models to calculate the threshold rainfall that can affect human walking, and the threshold rainfall of the concept of Grid to Grid (G2G) was calculated. In addition, although it was used a lot in the medical field in the past, a quantitative accuracy analysis was performed through the ROC analysis technique, which is widely used in natural phenomena such as drought or flood and machine learning. As a result of the analysis, the results of the time period similar to that of the actual and simulated immersion were obtained, and as a result of the ROC (Receiver Operating Characteristic) curve, the adequacy of the fair stage was secured with more than 0.7.
침수 흔적도 기반으로 도시침수 모형과 홍수추적모형의 침수양상 비교
최종화,전재현,김태형,김병식,Choi, Jonghwa,Jeon, Jaehyun,Kim, Taehyung,Kim, Byungsik 한국수자원학회 2021 한국수자원학회논문집 Vol.54 No.2
최근 이상강우에 따른 고빈도 강우 발생율의 증가 및 집중호우의 증가로 인한 침수발생 가능성이 증가하고 있다. 또한 도시화로 인한 인구집중과 개발 집중으로 인한 도시의 불투수층의 증가로 우수유출량이 증가하고 있다. 도시가 발달한 지역 특성상 하천 주변 및 저지대지역에 위치하고 있다. 내수침수분석에는 수리·수문분석을 바탕으로 우수관로 및 지표흐름 분석이 가능한 XP-SWMM 과 홍수-수문곡선과 강우-유출곡선을 추적 할 수 있는 FLO-2D 모형을 이용하여 실재 침수가 발생한 지역의 침수분석을 실시하여 비교 검토를 실시하였다. 두 모형의 침수발생 양상을 비교한 결과, 울진군 울진읍의 경우 LSSI 는 71.68%로 우수관망을 적용하지 않고 지형자료만을 이용한 FLO-2D도 양호하게 분석되었다. 따라서 XP-SWMM은 도시침수해소, 침수양상등 다양한 목적으로, FLO-2D는 침수양상만을 검토할 때 사용이 가능하다. In recent years, the possibility of flooding due to the increase in the incidence of high-frequency rainfall due to abnormal rainfall and the increase in concentrated torrential rain is increasing. Also, the amount of rainwater runoff is increasing due to the increase of the impermeable layer in the city due to the concentration of population due to urbanization and concentration of development. Due to the characteristics of the developed city, it is located in the vicinity of rivers and in the lowlands. For the analysis of inundation in water, using XP-SWMM, which can analyze stormwater pipelines and surface flows, and FLO-2D models that can track flood-sluice curves and rainfall-spill curves, based on hydraulic and hydrological analysis. Inundation analysis was conducted and comparative review was conducted. The patterns of flooding of the two models were compared, and a model suitable for domestic flooding was selected.
LSTM - MLP 인공신경망 앙상블을 이용한 장기 강우유출모의
안성욱,강동호,성장현,김병식,An, Sungwook,Kang, Dongho,Sung, Janghyun,Kim, Byungsik 한국수자원학회 2024 한국수자원학회논문집 Vol.57 No.2
Physical models, which are often used for water resource management, are difficult to build and operate with input data and may involve the subjective views of users. In recent years, research using data-driven models such as machine learning has been actively conducted to compensate for these problems in the field of water resources, and in this study, an artificial neural network was used to simulate long-term rainfall runoff in the Osipcheon watershed in Samcheok-si, Gangwon-do. For this purpose, three input data groups (meteorological observations, daily precipitation and potential evapotranspiration, and daily precipitation - potential evapotranspiration) were constructed from meteorological data, and the results of training the LSTM (Long Short-term Memory) artificial neural network model were compared and analyzed. As a result, the performance of LSTM-Model 1 using only meteorological observations was the highest, and six LSTM-MLP ensemble models with MLP artificial neural networks were built to simulate long-term runoff in the Fifty Thousand Watershed. The comparison between the LSTM and LSTM-MLP models showed that both models had generally similar results, but the MAE, MSE, and RMSE of LSTM-MLP were reduced compared to LSTM, especially in the low-flow part. As the results of LSTM-MLP show an improvement in the low-flow part, it is judged that in the future, in addition to the LSTM-MLP model, various ensemble models such as CNN can be used to build physical models and create sulfur curves in large basins that take a long time to run and unmeasured basins that lack input data.
변요셉(Yoseph Byun),김석천(Sukchun Kim),성주현(Joohyun Seong),천병식(Byungsik Chun),정혁상(Hyuksang Jung) 한국지반환경공학회 2014 한국지반환경공학회논문집 Vol.15 No.11
사면은 건설된 후에도 집중강우나 지진, 풍화 등 외부요인으로 인해 파괴가 발생할 수 있기 때문에, 사면의 안정적 유지관리를 위해서는 사면 붕괴의 가능성을 파악하는 것이 필요하다. 특히 암반사면은 암석의 취성적인 특성으로 인해 변위 계측 등과 같은 일반적인 방법으로는 파괴발생이전에 사전징후를 감지하기 매우 어렵다. 그러나 AE 기법을 사면에 적용한다면 변위가 발생하기 전에 파괴 시 발생된 AE 신호를 분석함으로써 일반적인 계측 방법보다 초기에 상태파악이 가능할 것이다. 본 논문에서는 한국의 암반사면 중 붕괴 이력을 가지고 있는 사면에 AE 기법을 적용하여 사면붕괴 가능성을 파악하였다. 그 결과 붕괴위험이 있는 사면에 AE 기법을 적용하면 사면의 위치별 붕괴 가능성을 가능할 것으로 판단된다. A slope may fail after construction owing to external factors such as localized rainfall, earthquake, and weathering. Therefore, the grasp of failure probability for slope failures is necessary to maintain their stability. In particular, it is very difficult to detect the symptoms of rock slope failure in advance by using traditional methods, such as displacement due to the brittleness of rocks. However, Acoustic Emission (AE) techniques can predict slope failures earlier than the traditional methods. This study grasped failure probability of slope by applying AE techniques to a rock slope with a history of collapse. When applying AE techniques to a slope that has a high probability of failure, the grasp of failure probability of the specific location became possible.