• 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
 
 
Master's Dissertation
DOI
https://doi.org/10.11606/D.3.2023.tde-21112023-110140
Document
Author
Full name
Thales Cesar Giriboni de Mello e Silva
E-mail
Institute/School/College
Knowledge Area
Date of Defense
Published
São Paulo, 2023
Supervisor
Committee
Corrêa, Pedro Luiz Pizzigatti (President)
Bittencourt, Túlio Nogueira
Ribeiro, Fernando Sgavioli
Title in Portuguese
Big Data aplicado à tarefa de categorização de níveis de severidade de vias férreas a partir de dados de veículos sensoriais.
Keywords in Portuguese
Aprendizado computacional
Big Data
Modelagem de dados
Transporte ferroviário
Vias permanentes
Abstract in Portuguese
A manutenção da condição de vias férreas é de fundamental importância para esse setor de transporte, responsável por considerável porção do fluxo de exportação da economia brasileira. Entretanto, eventuais falhas e defeitos acarretam interrupções da via, o que pode afetar negativamente o seu tráfego. Assim, convém utilizar o estado da arte de engenharia de computação para, a partir de dados coletados por diversos sensores acoplados a Vagões Instrumentados e Carros Controle, estimar a deterioração dos componentes de uma determinada ferrovia. Para tanto, técnicas de Big Data e Ciência dos Dados são utilizadas para captar, modelar e armazenar esses registros, permitindo o emprego de técnicas de aprendizado de máquina supervisionado e não-supervisionado para a definição de níveis de severidade da ferrovia, isto é, a quantificação do estado dos componentes da via permanente em níveis discretos de operacionalidade, onde a maior severidade indica falha iminente e a menor, condições normais de operação. Esses resultados são apresentados em uma plataforma web capaz de ser acessada remotamente, possibilitando o acompanhamento in loco das predições dos modelos computacionais.
Title in English
Untitled in english
Keywords in English
Big Data analytics
Machine learning
Predictive maintenance
Railtracks
Abstract in English
Railroad maintenance is of critical importance for that transportation sector, which is responsible for a significant portion of Brazilians exports. As such, occasional failures and defects can result in flow interruptions, which may negatively impact the output of the system. Therefore, it is desirable to use the state-of-the-art computer engineerings methods for developing a predictive algorithm for rail components deterioration estimation, using as input a variety of sensorial data gathered from instrumented ore wagons and track geometry control cars. For that goal, Big Data techniques are employed to capture, model and store these readings, paving the way for supervised and unsupervised machine learning algorithms to be used to train a model able to define a quantification of the rail component state into discrete operational levels, or severity levels for the railroads condition, where the greatest severity would indicate an imminent geometry defect, and the lowest would point to normal conditions. These results are then presented through a web application that can be remotely accessed, allowing for the in loco follow-up of the computational models predictions.
 
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.
Publishing Date
2023-11-22
 
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.