7th IMRMPT

Abstract

Abstract information

English title:	LSTM Networks To Predict Mechanical Behavior Of Additive Manufacturing Composites Reinforced With Carbon Fiber
Spanish title:	Redes LSTM para predecir el comportamiento mec�nico de materiales compuestos de fabricaci�n aditiva reforzados con fibra de carbono
English abstract:	Continuous fiber-reinforced additive manufacturing (cFRAM) composites have shown potential to improve the mechanical properties of 3D-printed polymers, but predicting their performance is a complex task due to their microstructure and various process parameters involved in their fabrication. Therefore, a long-short-term memory (LSTM) neural network is investigated to predict the mechanical properties, stiffness constants, and strength of cFRAM. In this study, the results of experimental tensile tests on continuous carbon fiber AM composites with variable volume fraction, printing direction, and fiber angle were used to train an LSTM neural network. The predictive model was compared to a micromechanical model of stiffness and strength to assess the accuracy, ease of implementation, and generalization capabilities of the machine learning algorithm. The results suggest that the prediction of the mechanical properties of cFRAM can be performed with reasonable accuracy.
Spanish abstract:	Los materiales compuestos de fabricaci�n aditiva reforzada con fibra continua (cFRAM) han demostrado potencial para mejorar las propiedades mec�nicas de los pol�meros impresos en 3D, pero predecir su desempe�o es una tarea compleja debido a su microestructura y varios par�metros de proceso involucrados en su fabricaci�n. Por lo tanto, se investiga una red neuronal de memoria a corto plazo (LSTM) para predecir las propiedades mec�nicas, las constantes de rigidez y la fuerza de cFRAM. En este estudio, se usaron los resultados de las pruebas experimentales de tracci�n en compuestos AM de fibra de carbono continuo con fracci�n de volumen variable, direcci�n de impresi�n y �ngulo de fibra para entrenar una red neuronal LSTM. El modelo predictivo se compar� con un modelo micromec�nico de rigidez y resistencia para evaluar la precisi�n, la facilidad de implementaci�n y las capacidades de generalizaci�n del algoritmo de aprendizaje autom�tico. Los resultados indican que la predicci�n de las propiedades mec�nicas de cFRAM se puede realizar con una precisi�n razonable.
Topic:
Presentation Type:
Speaker:

Authors

Name	Institution	Email
Octavio Andr�s Gonz�lez-Estrada	Universidad Industrial de Santander	agonzale@uis.edu.co
Edinson Lizarazo Sanchez	Universidad Industrial de Santander	edinson2228342@correo.uis.edu.co
Cristian Andr�s Hern�ndez Salazar	Universidad Industrial de Santander	cristian2228339@correo.uis.edu.co

Presentation

Room:	Principal Room
Date:	June 7, 2023
Hour:	09:00 - 09:20

Back

7^th IMRMPT
Seventh International Meeting for Researchers in Materials and Plasma Technology

Previous editions:

        1st IMRMPT 2011
        2nd IMRMPT 2013
        3rd IMRMPT 2015
        4th IMRMPT 2017
        5th IMRMPT 2019
        5+1 IMRMPT 2021

Contact information:
imrmpt@foristom.org

Copyright 2022 FORISTOM
All rights reserved