Author(s): Kevin D. Alvarez S.; Alejandro Mendoza; Moises Berezowsky
Linked Author(s):
Keywords: Hydromorphology; machine learning; multilayer perceptron; riverbed evolution; Telemac-2D
Abstract: Computing riverbed evolution is a fundamental part of hydromorphological analysis of rivers, as it identifies processes of erosion, sedimentation, displacement of riverbed features, and geometric alterations that impact the operation of hydraulic infrastructure and the management of natural channels. These processes are simulated using numerical models that solve the governing equations of flow and bed morphology. Solving these equations involves high computational costs, especially in large domains and over long periods. This research analyzes the capacity of Multilayer Perceptron (MLP) artificial neural networks as a tool for faster analysis of bed evolution processes. These networks are trained with results produced by 2D finite element models using TELEMAC-SISYPHE.
Year: 2026