Modelagem e metamodelagem de estrutura de material compósito laminado submetido ao regime de pós-flambagem
The structural stability of a design involving laminated composite materials is usually related to the determination of their critical buckling load. The present work aims to analyze the behavior of a composite plate, submitted to the post-buckling regime (load higher than the critical load). The co...
| Principais autores: | Siegel, Alison, Correa Neto, Pedro Arnaldo |
|---|---|
| Formato: | Trabalho de Conclusão de Curso (Graduação) |
| Idioma: | Português |
| Publicado em: |
Universidade Tecnológica Federal do Paraná
2020
|
| Assuntos: | |
| Acesso em linha: |
http://repositorio.utfpr.edu.br/jspui/handle/1/10259 |
| Tags: |
Adicionar Tag
Sem tags, seja o primeiro a adicionar uma tag!
|
| Resumo: |
The structural stability of a design involving laminated composite materials is usually related to the determination of their critical buckling load. The present work aims to analyze the behavior of a composite plate, submitted to the post-buckling regime (load higher than the critical load). The composite material structure is modeled using Abaqus® finite element software, which allows the analysis of the influence of various parameters, such as number of layers and their orientation angles, on the structural response. Subsequently, through the MATLAB® program and the data obtained through finite elements, a metamodel (neural networks) is used and trained to represent the composite structure, through its loads and displacements. The validity of the neural network metamodel is tested by presenting new input data, different from those used in training, and then comparing its response with that calculated by finite elements. For the present case study, it is concluded that neural networks have a partially satisfactory approximation of results. In other words, when training a neural network with the inputs of a plate and testing it with inputs of a different one with similar mechanical behavior (load versus displacement curve), we obtain satisfactory results. However, when testing the same neural network with inputs of a plate with different mechanical behavior, it does not present acceptable results. |
|---|