PSI - Issue 17

Available online at www.sciencedirect.com Structural Int grity Procedia 00 (2019) 000 – 000 Available online at www.sciencedirect.com ScienceDirect Structural Integrity Procedia 00 (2019) 000 – 000 Available online at www.sciencedirect.com ScienceDirect

www.elsevier.com/locate/procedia www.elsevier.com/locate/procedia

ScienceDirect

Procedia Structural Integrity 17 (2019) 698–703

ICSI 2019 The 3rd International Conference on Structural Integrity Seepage and dam deformation analyses with statistical models: support vector regression machine and random forest Ahmed Belmokre a, *, Mustapha Kamel Mihoubi a , David Santillan b a Laboratoire Mobilisation et Valorisation des Ressources en Eau (MVRE), Ecole Nationale Supérieure d’Hydraulique (ENSH), Blida , Algeria b Departamento de Ingeniería Civil :Hidráulica, Energía y Medio Ambiente, Universidad Politécnica de Madrid, Madrid, Spain. ICSI 2019 The 3rd International Conference on Structural Integrity Seepage and dam deformation analyses with statistical models: support vector regression machine and random forest Ahmed Belmokre a, *, Mustapha Kamel Mihoubi a , David Santillan b a Laboratoire Mobilisation et Valorisat on des Ressources n E u ( VRE), Ecole Nationale Supérieure d’Hy raulique (ENSH), Blida , Algeria b Departamento de Ingeniería Civil :Hidráulica, Energía y Medio Ambiente, Universidad Politécnica de Madrid, Madrid, Spain. Dam monitoring and their safety are an important concern of dam engineers. Seepage collected data are indicators of structure behavior, since seepage is influenced by environmental actions, such as air temperature, water temperature, and water level variation, and seepage flow rate is greatly influence by the presence of fractures. Consequently, the analysis of seepage collected data is an important monitoring task, as variations in the seepage can be the alarm for subsequent failures. Seepage data are widely analyzed with statistical models. In this work, we assess the performance of support vector regression machine and random forest models to predict seepage at different points in a case study and identify the most important environmental variables affecting flow rate. Dam monitoring and their safety are an important co cern of dam engineers. Seepage collected data are indicators of structure beh vi r, since seepage is influenced b environmental actions, such as air temperatur , water temperature, and water level v riation, and seepage flow rate is greatly influence by the resence of fractures. Consequently, the analysis of s epage collected data is an important monitoring task, as variations in the seepage can be the alarm for subsequent failures. Seepage data are widely analyzed with statistic l models. In this work, we assess the performance of support vector regression machine and random forest models to predict seepage at different points in a case study and identify the most important environmental variables affecting flow rate. Abstract Abstract

© 2019 The Authors. Published by Elsevier B.V. Peer-review under responsibility of the ICSI 2019 organizers. © 2019 The Authors. Published by Elsevier B.V. Peer-review under responsibility of the ICSI 2019 organizers. © 2019 The Authors. Published by Elsevier B.V. Peer-review under responsibility of the ICSI 2019 organizers.

Keywords: Dam monitoring, Seepage, Random forest, Support vector regression, Water temperature, Keywords: Dam monitoring, Seepage, Random forest, Support vector regression, Water temperature,

1. Introduction 1. Introduction

Seepage flow rate is an important performance indicator of the structural behavior of dams (Rubertis 2018) since it could provide an insight into physical changes in the structure of the dam. For that reason, the prediction and analysis of seepage recorded data at dams’ sites is an essential operation in dam monitoring tasks. Generally, it is done by deterministic models based on the finite element method or statistical methods (Santillan et al. 2011, 2014). Seepage flow rate is an important performance indicator of the structural behavior of dams (Rubertis 2018) since it could provide an insight into physical changes in the structure of the da . For that reason, the prediction and analysis of seepage recorded data at dams’ sites is an essential operation in dam monitoring tasks. Generally, it is done by deterministic models based on the finite element method or statistical methods (Santillan et al. 2011, 2014).

2452-3216 © 2019 The Authors. Published by Elsevier B.V. Peer-review under responsibility of the ICSI 2019 organizers. 2452-3216 © 2019 The Authors. Published by Elsevier B.V. Peer-review under responsibility of the ICSI 2019 organizers. * Corresponding author. E-mail address: a.belmokre@ensh.dz * Corresponding author. E-mail address: a.belmokre@ensh.dz

2452-3216  2019 The Authors. Published by Elsevier B.V. Peer-review under responsibility of the ICSI 2019 organizers. 10.1016/j.prostr.2019.08.093