• 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.3.2020.tde-31032021-162257
Document
Author
Full name
Arthur Henrique de Andrade Melani
E-mail
Institute/School/College
Knowledge Area
Date of Defense
Published
São Paulo, 2020
Supervisor
Committee
Souza, Gilberto Francisco Martha de (President)
Barretto, Marcos Ribeiro Pereira
Mauá, Denis Deratani
Pereira, Fabio Henrique
Silva, Leandro Dias da
Title in Portuguese
Diagnose de falhas em sistemas baseada em redes bayesianas e SysML.
Keywords in Portuguese
Engenharia (Sistemas; Diagnóstico)
Inferência bayesiana
Abstract in Portuguese
A crescente complexidade de equipamentos e sistemas, especialmente em áreas onde a segurança é de extrema importância, como a indústria da aviação ou usinas de energia nuclear, motivou a busca de métodos automatizados de diagnóstico de falhas. A diagnose de falhas representa o processo de identificação da origem de uma falha através de uma série de efeitos que ela causa no sistema ao qual pertence. O objetivo desta pesquisa é desenvolver um método para a diagnose de falhas baseada em um modelo do sistema em estudo através do uso de redes bayesianas em conjunto com a linguagem SysML. O método fornece um procedimento estruturado para a construção da rede bayesiana. A rede bayesiana obtida deve, por sua vez, apresentar os componentes que mais provavelmente são responsáveis por uma certa falha no sistema em estudo. A falha em questão será observada através da leitura de sensores presentes no sistema. O método proposto foi aplicado em uma usina a carvão, mais especificamente no sistema de dessulfurização de gases de combustão, e seus resultados foram comparados com o histórico de falhas da planta. A rede bayesiana obtida pelo método proposto mostrou-se capaz de diagnosticar falhas. Tanto o cálculo da probabilidade posterior quanto o algoritmo de máxima a posteriori (MAP) foram capazes de diagnosticar falhas no sistema FGD usando as evidências observadas pelos sensores que monitoram esse sistema.
Title in English
System fault diagnosis based on Bayesian networks and SysML.
Keywords in English
Bayesian network
Coal-fired power plant
Fault diagnosis
SysML
Abstract in English
The growing complexity of equipment and systems, especially in areas where safety is of the utmost importance, such as the aviation industry or nuclear power plants, has motivated the search for automated methods of fault diagnosis. Fault diagnosis represents the process of identifying the origin of a fault through a series of effects that it causes in the system to which it belongs. The objective of this research is to develop a method for fault diagnosis based on a model of the system under study through the use of Bayesian Networks in conjunction with the SysML language. The method provides a structured procedure for the construction of the Bayesian network. The Bayesian network obtained must, in turn, present the components that are most likely responsible for a certain failure of the system under study. The fault in question was observed through the reading of sensors present in the system. The proposed method is applied in a coal-fired power plant, more specifically in the flue gas desulfurization (FGD) system, and its results were compared with the plant failure history. The Bayesian network obtained through the proposed method proved to be capable of diagnosing faults. Both the posterior probability calculation and the maximum a posteriori (MAP) algorithm were able to diagnose faults in the FGD system using the evidence observed by the sensors that monitor that system.
 
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
2021-04-01
 
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.