• 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.85.2020.tde-03022020-110813
Document
Author
Full name
Davi Almeida Moraes
E-mail
Institute/School/College
Knowledge Area
Date of Defense
Published
São Paulo, 2019
Supervisor
Committee
Gonçalves, Iraci Martinez Pereira (President)
Aguiar, Paulo Marcos de
Bueno, Elaine Inacio
Mesquita, Roberto Navarro de
Silva, Antonio Teixeira e
Title in Portuguese
Planta experimental para monitoração e diagnóstico de falhas utilizando inteligência artificial
Keywords in Portuguese
digital twin
GMDH
inteligência artificial
planta experimental
redes neurais
Abstract in Portuguese
Neste trabalho foi desenvolvida uma planta experimental inspirada em um reator nuclear de potência do tipo PWR e posterior aplicação de Inteligência Artificial na Monitoração e Diagnóstico de Falhas, por meio dos métodos GMDH (Group Method of Data Handling) e RNA (Redes Neurais Artificiais). Com a planta experimental, tornou-se possível aplicar conceitos inovadores de modelagem de sistemas (Digital Twin) on line para a monitoração e diagnóstico de falhas individuais e/ou combinadas. Conclui-se que, embora ambos os sistemas de monitoração apresentaram resultados satisfatórios, o GMDH demonstrou um melhor desempenho em relação às Redes Neurais, pois além de apresentar valores de desvios médios menores do que o modelo utilizando Redes Neurais, foi possível realizar a monitoração de todas as variáveis, enquanto que utilizando Redes Neurais não foi possível monitorar as variáveis de potência do aquecedor, nível, e potência e vazões das bombas. A inserção de falhas em uma ou mais variáveis de temperatura, repercutiu na estimativa da rede para as demais variáveis, porém não impediu que o Sistema de Monitoração identificasse a falha. Para determinar o comportamento do Sistema de Monitoração com falhas múltiplas, foram aplicadas falhas simultâneas nos sensores de temperatura.
Title in English
Experimental plant for monitoring and fault diagnostic using artificial intelligence
Keywords in English
artificial intelligence
digital twin
experimental plants
GMDH
neural networks
Abstract in English
In this work we developed an experimental plant inspired by a PWR-type nuclear power reactor and applied later Artificial Intelligence in the Monitoring and Fault Diagnosis of Faults, through the Group Method of Data Handling (GMDH) and ANN (Artificial Neural Networks). Through the experimental plant, it became possible to apply innovative concepts of on-line system modeling (Digital Twin) for monitoring and diagnosing individual and / or combined faults. It is concluded that although both monitoring systems presented satisfactory results, GMDH showed a better performance compared to Neural Networks, because besides presenting smaller average deviation values than the model using Neural Networks, it was possible to perform the monitoring of all variables, while using Neural Networks, it was not possible to monitor the variables of heater power, level, and pump power and flow rates. The insertion of faults in one or more temperature variables interfered in the network estimation for the other variables, but did not prevent the Monitoring System from identifying the fault. To determine the behavior of the multiple fault monitoring system, simultaneous faults were applied to temperature sensors.
 
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.
2019MoraesPlanta.pdf (8.70 Mbytes)
Publishing Date
2020-02-07
 
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.