Prediction of Dam Seepage through a Machine Learning Technique and Its Application to Dam Diagnosis
OPEN ACCESS
Loading...
Author / Producer
Date
2024
Publication Type
Conference Paper
ETH Bibliography
no
Citations
Altmetric
OPEN ACCESS
Data
Rights / License
Abstract
This study proposes a machine learning model for dam management to achieve low-cost and rapid dam diagnosis based on observation data related to the seepage rate. This study employs extremely randomized trees (ERT), which is one of the machine learning techniques derived from the decision tree, as a learning algorithm. This relatively new algorithm is anticipated to yield an optimal outcome using a variety of observation information compared to other widely used tree-based approaches, such as the decision tree and random forest. In this study, transitionally observed seepage rates at actual rock-fill dams located in the Tohoku province in Japan were used to build the training and testing data. In addition to the measured seepage rate, 14 types of rainfall data including the rainfall from one day through 14 days before, the water level in the reservoir, and the temporal gradient of water level, were employed. This is because the seepage rate may be influenced not only by the rainfall event of the day but also by the status of the water level in the reservoir. The elapsed time since the floods test was also treated as label data in the learning process. These 44 variables were the data set in the learning model and were treated as features for seepage prediction. As a result, it was revealed that the water level in the reservoir, the rainfall on the day, and the temporal gradient of the water level are important for achieving highly accurate prediction.
Permanent link
Publication status
published
External links
Book title
Proceedings of the 10th International Symposium on Hydraulic Structures (ISHS 2024)
Journal / series
Volume
Pages / Article No.
567 - 575
Publisher
ETH Zurich
Event
10th International Symposium on Hydraulic Structures (ISHS 2024)
Edition / version
Methods
Software
Geographic location
Date collected
Date created
Subject
Dam Seepage; Features; Machine Learning; Soundness Assessment
Organisational unit
03820 - Boes, Robert / Boes, Robert
Notes
Funding
Related publications and datasets
Is part of: https://doi.org/10.3929/ethz-b-000675921