상세 보기
도시침수 대응을 위한 빗물펌프장 유입량 초단기 예측모델 연구
- 가순종;
- 박인선
초록
This study proposes a stormwater inflow prediction model aimed at maximizing the operational efficiency of rainwater pumping stations, which are critical facilities for urban flood prevention, in response to the increase in localized heavy rainfall driven by climate change and the expansion of impervious surfaces due to urbanization. The study area, the Dorimcheon basin, exhibits rapid runoff responses and short lag times due to its topographical characteristics, posing significant challenges for proactive pumping station management. To effectively capture these abrupt and nonlinear flow fluctuations, this research enhanced the Long Short-Term Memory (LSTM) model, specialized for time-series learning. First, to increase the precision of inflow data, dry-weather domestic sewage was statistically separated by time of day, and the dependent variables were refined through outlier treatment. Second, moving beyond the conventional geometric Thiessen polygon method that relied on a single station (AWS_410), the spatial scope was expanded to include three neighboring stations, optimizing input variables by assigning weights that reflect the basin's specific characteristics. Third, a Peak-Weighted Loss Function was introduced to improve the prediction accuracy of peak flows, which are vital for flood disaster prevention.The results demonstrated that the proposed model (Weight 2) achieved high predictive performance during the validation period, with a Mean Absolute Error (MAE) of 108.74 and a Correlation Coefficient (R) of 0.79. Notably, by incorporating a 10-minute lead time forecasting mechanism structured with a 20-minute lag shift (t-2), the Peak Flow Error (PFE) in high-flow sections was limited to 5.17%, significantly lower than that of the standard baseline model, indicating superior stability. The findings of this study are expected to contribute significantly to minimizing urban flood damage and enhancing energy efficiency by supporting decision-making in real-time rainwater pumping station operation systems.
키워드
- 제목
- 도시침수 대응을 위한 빗물펌프장 유입량 초단기 예측모델 연구
- 제목 (타언어)
- A Study on Ultra-Short-Term Inflow Forecasting Models for Rainwater Pumping Stations to Mitigate Urban Flooding
- 저자
- 가순종; 박인선
- 발행일
- 2026-09
- 유형
- Y
- 저널명
- 한국재난정보학회 논문집
- 권
- 23
- 호
- 3
- 페이지
- 761 ~ 770
- 언어
- KOR
- 출판사
- 한국재난정보학회
- 발행국가
- 대한민국
- 분량
- 10 페이지
- ISSN
- E 2671-5287
P 1976-2208