Simulation of seepage flow through embankment dam by using a novel extended Kalman filter based neural network paradigm: Case study of Fontaine Gazelles Dam, Algeria

Seepage flow through embankment dam is one of the most influential factors in failures of them. Thus, the monitoring and accurate measuring of seepage are crucial for the safety and construction cost of an embankment dam. In this study, an efficient data-intelligence paradigm comprised of Extended Kalman Filter integrated with the Feed Forward type Artificial Neural Network (EKF-ANN) scheme, as the main novelty, was developed for precise estimation of the daily seepage flow through embankment dam in Fontaine Gazelles Dam in Algeria. Here, three robust machine learning approaches, namely the Multilayer Perceptron (MLP) Neural Networks, Radial Basis Function-Neural Networks (RBF-NN), and Random Forest (RF), were examined for evaluating the capability of the EKF-ANN in the prediction of seepage flow. According to the obtained results, the EKF-ANN paradigm outperformed the MLP, RF, and RBF-NN, respectively. Besides, the leverage approach was applied to report the applicability domain of provided models.

Automated summary

In the study on seepage flow through Fontaine Gazelles Dam in Algeria, an efficient data-intelligence paradigm, EKF-ANN, was developed for precise estimation of daily seepage flow. The EKF-ANN paradigm outperformed other machine learning approaches like MLP, RF, and RBF-NN in predicting seepage flow. Leveraging approaches were also used to determine the applicability domain of the models.