Título: Detecção de Falha por Redes Neurais em uma Viga Flexível
Autores: Alves Jr., Marco Antonio O.; Nóbrega, Eurípedes G.; Pavanello, Renato
Resumo: The main goal of this work is to present a neural network configuration able to detect faults in an active controlled mechanical plant. Because it was built as an output estimator, the developed method is here called a neural observer fault detector. The neural observer results for fault detection are compared to the results of an observer-based known method. Concentrated mass variation on a flexible beam is the fault considered for simulated and experimental cases. Their effects were analyzed by abrupt addition of a mass on three different positions on the beam, causing variation on the vibration signals. A finite element model was developed for the simulation of the fault, and PZT actuators and sensors were used in the experimental setup. The results demonstrate the applicability of the method, when the plant vibration is free of control and also for the controlled case. The advantages of this approach are that a mathematical model for the plant is not necessary and that the method may be applied to nonlinear systems as well.
Código DOI: 10.21528/CBRN2001-013
Artigo em pdf: 5cbrn_013.pdf
Arquivo BibTex: 5cbrn_013.bib