DOI QR코드

DOI QR Code

Multidimensional data generation of water distribution systems using adversarially trained autoencoder

적대적 학습 기반 오토인코더(ATAE)를 이용한 다차원 상수도관망 데이터 생성

  • Kim, Sehyeong (Department of Civil, Environmental and Architectural Engineering, Korea University) ;
  • Jun, Sanghoon (Hyper-converged Forensic Research Center for Infrastructure, Korea University) ;
  • Jung, Donghwi (School of Civil, Environmental and Architectural Engineering, Korea University)
  • 김세형 (고려대학교 건축사회환경공학과) ;
  • 전상훈 (고려대학교 초융합건설포렌식연구센터) ;
  • 정동휘 (고려대학교 건축사회환경공학부)
  • Received : 2023.05.19
  • Accepted : 2023.07.21
  • Published : 2023.07.31

Abstract

Recent advancements in data measuring technology have facilitated the installation of various sensors, such as pressure meters and flow meters, to effectively assess the real-time conditions of water distribution systems (WDSs). However, as cities expand extensively, the factors that impact the reliability of measurements have become increasingly diverse. In particular, demand data, one of the most significant hydraulic variable in WDS, is challenging to be measured directly and is prone to missing values, making the development of accurate data generation models more important. Therefore, this paper proposes an adversarially trained autoencoder (ATAE) model based on generative deep learning techniques to accurately estimate demand data in WDSs. The proposed model utilizes two neural networks: a generative network and a discriminative network. The generative network generates demand data using the information provided from the measured pressure data, while the discriminative network evaluates the generated demand outputs and provides feedback to the generator to learn the distinctive features of the data. To validate its performance, the ATAE model is applied to a real distribution system in Austin, Texas, USA. The study analyzes the impact of data uncertainty by calculating the accuracy of ATAE's prediction results for varying levels of uncertainty in the demand and the pressure time series data. Additionally, the model's performance is evaluated by comparing the results for different data collection periods (low, average, and high demand hours) to assess its ability to generate demand data based on water consumption levels.

최근 계측 기술의 발전으로 압력계와 유량계 등 다양한 센서를 설치하여 상수도관망의 상태를 효과적으로 파악할 수 있게 되었으나, 도시가 광범위하게 개발됨에 따라 계측 신뢰도에 영향을 미치는 변수는 다양해지고 있다. 특히 상수도관망 분석에 중요한 영향력을 가지는 수요 데이터의 경우 직접 계측의 난이도가 높고 결측이 발생하기 쉬운 것으로 알려져 데이터 생성의 중요도가 증가하고 있다. 본 논문에서는 상수도관망에서 누락된 데이터를 정확하게 생성하기 위해 생성적 딥러닝 모델에 기반한 적대적 학습 기반 오토인코더(ATAE) 모델을 제안한다. 제안된 모델은 판별 신경망과 생성 신경망의 두 가지 신경망의 적대적 학습을 사용하여 압력 데이터로부터 수요 데이터를 생성한다. 학습이 완료된 ATAE 모델의 생성 신경망은 관망의 계측되는 압력 데이터가 존재하는 경우, 그로부터 추정된 관망 수요 데이터를 제공할 수 있다. ATAE 모델은 미국 텍사스주 오스틴의 실제 상수도망에 적용되어 성능이 검증되었다. 수요 및 압력 시계열 데이터의 불확실성 정도에 따른 ATAE 예측 결과의 정확도를 비교하여 데이터 불확실성의 영향을 분석하였으며, 또한 수요 수준에 따른 데이터 수집 기간별 생성 결과를 비교하여 이에 따른 데이터 생성 성능을 검토하였다.

Keywords

Acknowledgement

이 성과는 정부(과학기술정보통신부)의 재원으로 한국연구재단의 지원을 받아 수행된 연구임(No. 2020R1C1C1006481).

References

  1. Bagheri, A., Monsef, H., and Lesani, H. (2015). "Integrated distribution network expansion planning incorporating distributed generation considering uncertainties, reliability, and operational conditions." International Journal of Electrical Power & Energy Systems, Vol. 73, pp. 56-70. https://doi.org/10.1016/j.ijepes.2015.03.010
  2. Bragalli, C., Neri, M., and Toth, E. (2019). "Effectiveness of smart meter-based urban water loss assessment in a real network with synchronous and incomplete readings." Environmental Modelling & Software, Vol. 112, pp. 128-142. https://doi.org/10.1016/j.envsoft.2018.10.010
  3. Brentan, B.M., Meirelles, G.L., Manzi, D., and Luvizotto, E. (2018). "Water demand time series generation for distribution network modeling and water demand forecasting." Urban Water Journal, Vol. 15, No. 2, 150-158. https://doi.org/10.1080/1573062X.2018.1424211
  4. Goodfellow, I.J., Pouget-Abadie, J., Mirza, M., Xu, B., Warde-Farley, D., Ozair, S., and Bengio, Y. (2014). "Generative adversarial networks, 1-9." arXivpreprint, arXiv:1406.2661.
  5. Jang, H., Jung, D., and Jun, S. (2022). "Comparison of ANN model's prediction performance according to the level of data uncertainty in water distribution network." Journal of Korea Water Resources Association, Vol. 55, No. 12, pp. 1295-1303. https://doi.org/10.3741/JKWRA.2022.55.S-1.1295
  6. Jun, S., Jung, D., and Lansey, K.E. (2021). "Comparison of imputation methods for end-user demands in water distribution systems." Journal of Water Resources Planning and Management, Vol. 147, No. 12, 04021080.
  7. Jung, D., Chung, G., and Kim, J.H. (2010). Optimal design of water distribution systems considering uncertainties in demands and roughness coefficients. Water Distribution Systems Analysis 2010, In 12th Annual Conference, Tucson, AZ, U.S., pp. 1390-1399.
  8. Kabir, G., Tesfamariam, S., Hemsing, J., and Sadiq, R. (2020). "Handling incomplete and missing data in water network database using imputation methods." Sustainable and Resilient Infrastructure, Vol. 5, No. 6, pp. 365-377. https://doi.org/10.1080/23789689.2019.1600960
  9. Karras, T., Laine, S., and Aila, T. (2019). "A style-based generator architecture for generative adversarial networks." In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 4401-4410.
  10. Mirza, M., and Osindero, S. (2014). "Conditional generative adversarial nets." arXivpreprint, arXiv:1411.1784.
  11. Rodriguez, R., Pastorini, M., Etcheverry, L., Chreties, C., Fossati, M., Castro, A., and Gorgoglione, A. (2021). "Water-quality data imputation with a high percentage of missing values: A machine learning approach." Sustainability, Vol. 13, No. 11, 6318.
  12. Rombach, R., Blattmann, A., Lorenz, D., Esser, P., and Ommer, B. (2022). "Highresolution image synthesis with latent diffusion models." In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, New Orleans, LA, U.S., pp. 10684-10695.
  13. Shrivastava, A., Pfister, T., Tuzel, O., Susskind, J., Wang, W., and Webb, R. (2017). "Learning from simulated and unsupervised images through adversarial training." In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Honolulu, HI, U.S., pp. 2107-2116.
  14. Sun, S., Yeh, C.F., Hwang, M.Y., Ostendorf, M., and Xie, L. (2018). Domain adversarial training for accented speech recognition. In 2018 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), IEEE, Calgary, Canada, pp. 4854-4858.
  15. Zanfei, A., Menapace, A., Brentan, B.M., and Righetti, M. (2022). "How does missing data imputation affect the forecasting of urban water demand?." Journal of Water Resources Planning and Management, Vol. 148, No. 11, 04022060.