All Issue

2026 Vol.22, Issue 3 Preview Page

Original Article

A Study on Ultra-Short-Term Inflow Forecasting Models for Rainwater Pumping Stations to Mitigate Urban Flooding

도시침수 대응을 위한 빗물펌프장 유입량 초단기 예측모델 연구

가순종1ㆍ박인선2*

Soon Jong Ka1, In Seon Park2*

1Graduate Student, Department of ICT Convergence Safety, The Graduate School of Chung-Ang University, Seoul, Korea
2Professor, The Head of Department of ICT Convergence Safety Department, The Graduate School of Chung-Ang University, Seoul, Korea
30 September 2026. pp. 761-770
PDF Citation
Abstract
Purpose: This study proposes an inflow prediction model to maximize the operational efficiency of rainwater pumping stations, which are critical facilities for urban flood prevention, in response to intensified localized heavy rainfall and expanding impervious surfaces. The Dorimcheon basin, the study area, shows rapid runoff responses and short lag times, making proactive pump operation difficult. Method: The Long Short-Term Memory (LSTM) model, specialized for time-series learning, was enhanced in three ways: dry-weather domestic sewage was statistically separated by time of day and outliers were treated to refine the dependent variable; based on Thiessen polygon analysis, a 50% spatial weight was assigned to the dominant station (AWS_410) among four stations; and a Peak-Weighted Loss Function was introduced to improve peak-flow prediction. Result: The proposed model (Wp=2.0) achieved an MAE of 108.74 m3/10min and a correlation coefficient (R) of 0.789 during validation. With a 20-minute lag shift (t-2) and a 10-minute lead-time forecasting structure, the Peak Flow Error (PFE) in high-flow sections was limited to 5.17%, far lower than that of the baseline model. Conclusion: The findings can support real-time decision-making in pumping station operation, minimizing urban flood damage and enhancing energy efficiency.
Keywords
Rainwater Pumping Station
Inflow Forecasting
LSTM
Spatial Weighting Design
Peak-Weighted Loss Function
연구목적: 본 연구는 국지성 집중호우의 증가와 도시화에 따른 불투수면 확산에 대응하여, 도심 침수 방지의 핵심 시설인 빗물펌프장의 운영 효율을 극대화하기 위한 유입량 예측 모형을 제안한다. 대상지인 도림천 유역은 강우 시 유출 응답이 매우 빠르고 지체 시간이 짧아 선제적 펌프장 운영에 어려움이 있다. 연구방법: 시계열 학습에 특화된 LSTM 모형을 다음과 같이 개선하였다. 첫째, 기저 생활오수량을 시간대별로 분리하고 이상치를 처리하여 종속변수를 고도화하였다. 둘째, 기하학적 티센 분석을 토대로 주관측소(AWS_410)에 50%의 공간 가중치를 부여하고 4개 관측소로 공간적 범위를 확대하였다. 셋째, 피크 가중치 손실함수를 도입하여 첨두유량 예측 정확도를 높였다. 연구결과: 최종 제안 모델(Wp=2.0)은 검증 기간 MAE 108.74m3/10min, 상관계수(R) 0.789의 우수한 예측성능을 보였고, 유역 도달시간을 반영한 강우자료 20분 시프트(t-2)와 10분 선행예측(t+1) 구조를 통해 고유량 구간 첨두유량오차율(PFE)이 5.17%로 낮게 나타나 높은 안정성을 확보하였다. 결론: 본 연구의 성과는 실시간 빗물펌프장 운영시스템의 의사결정을 지원함으로써 침수피해 최소화와 에너지 효율 향상에 기여할 것으로 기대된다.
키워드
빗물펌프장
유입량 예측
LSTM
공간 가중치 설계
피크 가중치 손실함수
References
  1. Choi, H.-S., Yoon, S.-K. (2021). “Deep learning-based conduit water level prediction for Shinwol underground stormwater tunnel operation.” In Proceedings of the Korea Water Resources Association Academic Conference, Gwangju, Korea. p. 418.
  2. Hajgató, G., Paál, G., Gyires-Tóth, B. (2020). “Deep reinforcement learning for real-time optimization of pumps in water distribution systems.” Journal of Water Resources Planning and Management, Vol.146, No.11, 04020079. 10.1061/(ASCE)WR.1943-5452.0001287
  3. Huang, Z., Wang, Y., Dong, X. (2025). “Dimensions of superiority: How deep reinforcement learning excels in urban drainage system real-time control.” Water Research X, Vol. 28, 100313. 10.1016/j.wroa.2025.100313 40007797 PMC11850747
  4. Jo, D.-J. (2007). A GA-Fuzzy Model for the Optimal Control of Interior Drainage System. Ph.D. Dissertation, Korea University.
  5. Joo, J.-G., Jeong, I.-S., Kang, S.-H. (2024). “Deep reinforcement learning for multi-objective real-time pump operation in rainwater pumping stations.” Water, Vol.16, No. 23, 3398. 10.3390/w16233398
  6. Joo, J.-G., Yang, D.-G., Yoo, J.-M., Jung, D.-H., Kim, J.-H. (2010). “Improvement and application of pump station operating system and economic analysis of the application.” Journal of the Korean Society of Hazard Mitigation, Vol. 10, No. 3, pp. 155-165.
  7. Jun, H., Lee, Y.-J., Kim, J.-H., Lee, J.-H. (2007). “Development of a real time control model for urban drainage systems.” Journal of Korean Society on Water Environment, Vol. 23, No. 5, pp. 748-755.
  8. Kim, D.-H., Kim, S.-I., Koh, J.-H. (2023). “A Study on a Flow Rate Prediction CNN-LSTM Model Based on Deep Learning using Meteorological Data.” Journal of the Korea Academia-Industrial cooperation Society, Vol. 24, No. 8, pp. 22-31. 10.5762/KAIS.2023.24.8.22
  9. Kim, H., Jung, D. (2025). “Development of a spatio-temporal deep learning model for urban drainage outflow prediction.” Journal of Korea Water Resources Association, Vol. 58, No. 10, pp. 927-937. 10.3741/JKWRA.2025.58.10.927
  10. Kim, H.-I., Han, K.-Y., Lee, J.-Y. (2020). “Prediction of urban flood extent by LSTM model and logistic regression.” KSCE Journal of Civil and Environmental Engineering Research, Vol. 40, No. 3, pp. 273-283.
  11. Kim, Y.-N., Lee, E.-H. (2021). “Development of Automatic Operation Considering Gate Operation for Drainage Pump Station and Resilience Index Based on Subcatchment Runoff.” Journal of the Korea Academia-Industrial cooperation Society, Vol. 22, No. 12, pp. 427-436. 10.5762/KAIS.2021.22.12.427
  12. Kow, P.-Y., Liou, J.-Y., Yang, M.-T., Lee, M.-H., Chang, L.-C., Chang, F.-J. (2024). “Advancing climate-resilient flood mitigation: Utilizing transformer-LSTM for water level forecasting at pumping stations.” Science of the Total Environment, Vol. 927, 172246. 10.1016/j.scitotenv.2024.172246
  13. Lee, E.-H. (2023). “Inflow prediction of centralized reservoir for the operation of pump station in urban drainage systems using improved multilayer perceptron using existing optimizers combined with metaheuristic optimization algorithms.” Water, Vol. 15, No. 8, 1543. 10.3390/w15081543
  14. Lee, J.-H., Lee, Y.-J., Kim, J.-H., Jun, H.-D. (2007). “Development of a pump operation rule in a drainage pump station using a real time control model for urban drainage system.” Journal of Korea Water Resources Association, Vol. 40, No. 11, pp. 877-886. 10.3741/JKWRA.2007.40.11.877
  15. Neagoe, A., Tică, E.-I., Vuță, L.-I., Nedelcu, O., Dumitran, G.-E., Popa, B. (2025). “Hybrid LSTM-ARIMA model for improving multi-step inflow forecasting in a reservoir.” Water, Vol. 17, No. 21, 3051. 10.3390/w17213051
  16. Park, S.-H. (2019). Prediction of Water Level in Urban Area Using Deep Learning : A Case Study of Dorim Basin. Master Thesis, University of Seoul.
  17. Seoul Metropolitan Government (2020). Establishment of Smart Rainfall Analysis System for Rainwater Pumping Stations in Seoul. Water Circulation Safety Bureau, Seoul Metropolitan Government.
  18. Sim, K.-B., Yoo, E.-S. (2018). “Determination of optimal operation water level of rain water pump station using optimization technique.” Journal of the Korea Academia-Industrial Cooperation Society, Vol. 19, No. 7, pp. 337-342. 10.5762/KAIS.2018.19.7.337
  19. Sunela, M.I. (2017). Real-Time Control Optimization of Water Distribution System with Storage. Ph.D. Dissertation, Tallinn University of Technology, Tallinn, Estonia.
Information
  • Publisher :The Korean Society of Disaster Information
  • Publisher(Ko) :한국재난정보학회
  • Journal Title :Journal of the Society of Disaster Information
  • Journal Title(Ko) :한국재난정보학회논문집
  • Volume : 22
  • No :3
  • Pages :761-770
  • DOI :https://doi.org/10.15683/kosdi.2026.9.30.761
Journal Informaiton Journal of the Society of Disaster Information Journal of the Society of Disaster Information
  • NRF
  • KISTI Current Status
  • KISTI Cited-by
  • crosscheck
  • orcid
  • open access
  • ccl
Journal Informaiton Journal Informaiton - close