Estimativa do índice de irregularidade internacional a partir de sinais obtidos por sensores de smartphones embarcados em veículos
In Brazil, the transport of cargo and passengers takes place mainly through highways. In order to provide a satisfactory bearing quality, it is essential that the country has assessment instruments that allow constant monitoring of this infrastructure. In recent years, smartphones can be used as eva...
| Autor principal: | Freitas, Natanael Evangelista de |
|---|---|
| Formato: | Trabalho de Conclusão de Curso (Graduação) |
| Idioma: | Português |
| Publicado em: |
Universidade Tecnológica Federal do Paraná
2022
|
| Assuntos: | |
| Acesso em linha: |
http://repositorio.utfpr.edu.br/jspui/handle/1/28610 |
| Tags: |
Adicionar Tag
Sem tags, seja o primeiro a adicionar uma tag!
|
| Resumo: |
In Brazil, the transport of cargo and passengers takes place mainly through highways. In order to provide a satisfactory bearing quality, it is essential that the country has assessment instruments that allow constant monitoring of this infrastructure. In recent years, smartphones can be used as evaluation instruments because their sensors capture the vertical acceleration signals that can be related to the International Roughness Index (IRI), the most used objective parameter for the characterization of the bearing quality of floors. Therefore, the objective of this work was to develop intelligent models to relate the vertical acceleration signal to the International Roughness Index. The models developed are artificial neural networks (ANN) namely: convolutional neural networks and recurrent neural networks. The analysis of the results of this work considers the result of a simple linear regression presented in works found in the literature and is used for comparison with the result obtained with the system developed in this work of neural networks. When comparing the results, it is noted that the neural networks show a significant improvement in the correlation between the International Roughness Index and vertical acceleration, proving to be an excellent alternative for solving this problem. TNeural networks obtained a coefficient of determination above 0.95. |
|---|