Examinando por Materia "Redes neurales"

Mostrando 1 - 2 de 2
  • Miniatura
    PublicaciónAcceso abierto
    Una aproximación práctica a las redes neuronales artificiales.
    (2017-08-28) Caicedo Bravo, Eduardo Francisco; López Sotelo, Jesús Alfonso
    CONTENIDO: Generalidades sobre redes neuronales artificiales -- Redes neuronales perceptron y adaline -- Perceptron multicapa y algoritmo backpropagation -- Red neuronal de Hopfield -- Mapas auto-organizados de Kohonen -- Red neuronal de Base radial (RBF).
  • Miniatura
    PublicaciónAcceso abierto
    Geometry optimization of a francis turbine for efficiency, cavitation and erosion using computational fluid dynamics
    (Universidad del Valle, 2018) Aponte Núñez, Rubén Darío; Rodríguez Pulecio, Sara Aida; Ladino Ospina, Jair Alexander
    In recent years, the application of numerical computational models based on computational fluid dynamics (CFD) to industrial problems has been increasing; Today CFD is used to optimize and develop equipment and processes in many types of industry including the energy industry. The main advantage of the solutions with CFD is in the obtaining of the operating conditions and the analysis of internal and external flows, which experimentally is very difficult and expensive to achieve. This document presents the results of the research project of a master's degree in engineering with an emphasis in mechanical engineering where the geometry that minimizes the erosive wear by hard particle and cavitation for the different operating regimes maintaining the efficiency of the 10MW Francis turbine of the Amaime hydroelectric plant was obtained. To achieve this, a Simplified Virtual Laboratory (SVL) methodology was implemented, consisting of the use of Computational Fluid Dynamics and an optimization technique. First, the simulation of the current geometry of the turbine was carried out to characterize and verify, with experimental data, that the model represents the current operating conditions; this required to generate 3D CAD geometries by means of planes and reverse engineering using three-dimensional scanning of complex elements of the turbine such as blades. Second, it was required to optimize the geometry of the runner blades, guide vanes, covers and labyrinths by the combined use of factorial design of experiments, artificial neural networks (ANN) and optimization techniques by genetic algorithms.