Implementación de redes neuronales para la predicción de la capacidad de absorción de energía de concreto de alto desempeño sometido a ensayo de tracción directa

The following directed work investigates the efficiency of the implementation of artificial neural networks (ANN) for the prediction of the energy absorption capacity (g) of ultra-high-performance fiber-reinforced concrete (UHPFRC) based on its dosage, subjected to direct traction. For the development of the project, 500 dosages of UHPFRC concretes compiled from the scientific literature will be used in order to adjust the mathematical model. To improve the model, the collected data was divided into training and testing data. The network was fitted using k-fold validation with the training data and evaluated with the test data. For the model, the dosages of UHPFRC reinforced with a fiber or with a hybrid mixture of two fibers were considered, from a wide range of fibers, such as straight steel fibers, hook-finished steel fibers, twisted steel fibers, PVA, polyethylene fibers and polypropylene fibers. In addition, an experimental validation of the network was carried out. The results demonstrated the efficiency of the model according to the statistical parameters used, as well as its precision and versatility to treat new data.

• Autores: Rojas Grillo, Julian Santiago
• Fecha de publicación: 2020
• Institución: Escuela Colombiana de Ingeniería Julio Garavito
• Repositorio: Repositorio Institucional ECI
• Idioma: spa
• OAI Identifier: oai:repositorio.escuelaing.edu.co:001/1338
• Acceso en línea: https://repositorio.escuelaing.edu.co/handle/001/1338
• Palabra clave: UHPFRC; Ensayo de tracción directa; ANN; Modelación; Capacidad de absorción de energía; UHPFRC; Direct tensile test; ANN; Modeling; Energy absorption capacity
• License: http://purl.org/coar/access_right/c_abf2