• JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
 
  Bookmark and Share
 
 
Doctoral Thesis
DOI
https://doi.org/10.11606/T.18.2018.tde-14032018-114100
Document
Author
Full name
José Remo Ferreira Brega
Institute/School/College
Knowledge Area
Date of Defense
Published
São Carlos, 1997
Supervisor
Committee
Sória, Manoel Henrique Alba (President)
Domingues, Felippe Augusto Aranha
Fabbri, Glauco Tulio Pessa
Kawamoto, Eiji
Suzuki, Carlos Yukio
Title in Portuguese
A utilização de redes neurais artificiais em um sistema de gerência de pavimentos
Keywords in Portuguese
Redes neurais artificiais
Sistemas de gerência de pavimentos
Abstract in Portuguese
Esta tese apresenta o estudo para utilização de Redes Neurais Artificiais para avaliar o estado do pavimento e apoiar as decisões dentro de um Sistema de Gerência de Pavimentos. É apresentado um método para a avaliação da condição em pavimentos flexíveis, utilizando redes neurais MLP backpropagation. Neste caso para a extração das características dos pavimentos são utilizados dois métodos muito empregados pelos órgãos rodoviários: o "Índice de Gravidade Global" e a "irregularidade". Os experimentos demonstraram que as redes neurais simulam satisfatoriamente o estado dos pavimentos. Para se verificar a possibilidade de utilização em outros problemas, o processo foi empregado para o projeto de restauração de pavimentos flexíveis. Foi utilizada a DNER-PRO 159/85 para a extração das características dos pavimentos. Os experimentos demonstraram que as redes neurais também simulam convenientemente as características do pavimento. Como exemplo de ferramenta de apoio à gerência foi desenvolvido um protótipo computacional em ambiente gráfico, onde o critério de decisão baseia-se nas redes neurais estudadas. São descritas todas as suas funções e forma de funcionamento.
Title in English
Use of artificial neural networks in a management pavement system
Keywords in English
Artificial neural networks
Management pavement system
Abstract in English
A study of artificial neural networks for evaluating the pavement condition and for supporting decisions within a Pavement Management System is presented. The method for condition evaluation of flexible pavements using the MLP backpropagation technique is described. Two of the most used procedures for detecting the pavement conditions were applied: the "overall severity index" (Brazilian IGG) and the "irregularity index". The experiments demonstrated that the neural networks satisfactorily simulated the state of the pavement. In order to test the applications in other problems, the method was used for pavement overlay design through neural network, using the same MLP backpropagation technique. For detecting the pavement conditions the DNER-PRO 159/85 was applied. Tests with the model also demonstrated that the neural networks appropriately simulate the pavement characteristics. A computational prototype developed in a graphical computer environment, where the decision criteria are based on the neural networks studied, is presented as an example. All the functions and working details of such prototype are described.
 
WARNING - Viewing this document is conditioned on your acceptance of the following terms of use:
This document is only for private use for research and teaching activities. Reproduction for commercial use is forbidden. This rights cover the whole data about this document as well as its contents. Any uses or copies of this document in whole or in part must include the author's name.
Tese_Brega_JoseRF.pdf (41.31 Mbytes)
Publishing Date
2018-03-16
 
WARNING: Learn what derived works are clicking here.
All rights of the thesis/dissertation are from the authors
CeTI-SC/STI
Digital Library of Theses and Dissertations of USP. Copyright © 2001-2024. All rights reserved.