You are in the accessibility menu

Please use this identifier to cite or link to this item: http://acervodigital.unesp.br/handle/11449/87204
Full metadata record
DC FieldValueLanguage
dc.contributor.advisorAguiar, Paulo Roberto de [UNESP]-
dc.contributor.authorMartins, Cesar Henrique Rossinoli-
dc.date.accessioned2014-06-11T19:22:34Z-
dc.date.accessioned2016-10-25T18:57:09Z-
dc.date.available2014-06-11T19:22:34Z-
dc.date.available2016-10-25T18:57:09Z-
dc.date.issued2013-03-20-
dc.identifier.citationMARTINS, Cesar Henrique Rossinoli. Classificação de níveis de desgaste de dressadores de ponta única utilizando sinais de emissão acústica e redes neurais artificiais. 2013. 109 p. Dissertação (mestrado) - Universidade Estadual Paulista Júlio de Mesquita Filho, Faculdade de Engenharia, 2013.-
dc.identifier.urihttp://hdl.handle.net/11449/87204-
dc.identifier.urihttp://acervodigital.unesp.br/handle/11449/87204-
dc.description.abstractA retificação é um processo de acabamento de peças, que objetiva produtos e superfícies avançadas. Porém, como o constante atrito entre a peça e rebolo, este perde a agressividade, de modo que o resultado da retificação fica prejudicado. Quando isso ocorre é imprescindível à realização do processo de dressagem, que consiste em reavivar os grãos gastos do rebolo. Como as condições de dressagem provocam uma grande influência no desempenho da operação de retificação, monitorá-las durante todo o processo pode aumentar a sua eficiência. Assim, no presente trabalho foi realizado um estudo do desgaste de três tipos de diamantes para dressadores e, posteriormente, foram desenvolvidos modelos neurais, baseados em redes MLP e de Kohonen, capazes de classificar o nível de desgaste dos dressadores. Para se atingir esse objetivo foram ensaiados três tipos de dressadores de ponta única, um com diamante sintético CVD e dois com diamantes naturais, Mato Grosso e Brasil Extra. Serviram de entradas para os modelos neurais as estatísticas RMS e ROP obtidas após o estudo do conteúdo harmônico do sinal da emissão acústica. Para cada diamante foi obtido um modelo neural mais apto para as características do diamante. Os resultados mostraram um bom desempenho das redes neurais empregadas, atingindo-se taxas de acerto de 100% em alguns modelos neuraispt
dc.description.abstractGrinding is a part finishing process for advanced products and surfaces. However, continuous friction between the work piece and the grinding wheel causes the latter to lose its sharpness, thus impariring the grinding results. This is when the dressing process is required, which consist of sharpening the worn grains of the grinding wheel. The dressing conditions strongly affect the performance of the grinding operation; hence, monitoring them thoughout the process can increase its efficiency. The main objective of this study was to classify the levels of wear of a single-point dresser during the operation; hence monitoring them throughout the process can increase its efficiency. The main objective of this study was to classify the levels of wear of a single-point dresser during the operation using neural models. A digital signal processing of acoustiv emission was used to obtain the network inputs. In the experiments three types of single-point dresser were used, one with CVD synthetic diamond and two with natural diamonds, Mato Grosso and Brasil Extra. The harmonic content of the acoustic emission signal was found to be influenced by the condition of the dresser, and when used to feed the neural model it is possible to classify the wear level. The results showed a good performance of the neural networks employed, reaching a hit rate of 100% for some modelsen
dc.format.extent109 p. : il.-
dc.language.isopor-
dc.publisherUniversidade Estadual Paulista (UNESP)-
dc.sourceAleph-
dc.subjectPeças de maquinas - Acabamentopt
dc.subjectRetificação e polimentopt
dc.subjectEmissão acústicapt
dc.subjectRedes neurais (Computação)pt
dc.subjectRebolospt
dc.subjectMachine parts - Finishes and finishingpt
dc.titleClassificação de níveis de desgaste de dressadores de ponta única utilizando sinais de emissão acústica e redes neurais artificiaispt
dc.typeoutro-
dc.contributor.institutionUniversidade Estadual Paulista (UNESP)-
dc.rights.accessRightsAcesso aberto-
dc.identifier.filemartins_chr_me_bauru.pdf-
dc.identifier.aleph000718870-
dc.identifier.capes33004056087P2-
Appears in Collections:Artigos, TCCs, Teses e Dissertações da Unesp

There are no files associated with this item.
 

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.