Análise de componentes independentes aplicadas em compressores herméticos alternativo para mitigação das fontes de ruído

• Ano de defesa: 2017
• Autor(a) principal: Fagundes Neto, Marlipe Garcia
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Uberlândia; Brasil; Programa de Pós-graduação em Engenharia Mecânica
• Palavras-chave em Português: Análise de componentes independentes; Compressores herméticos alternativo; Fontes de ruído; Independent component analysis; Reciprocating hermetic compressors; Noise source; CNPQ::ENGENHARIAS::ENGENHARIA MECANICA::PROJETOS DE MAQUINAS
• Link de acesso: https://repositorio.ufu.br/handle/123456789/20848; http://dx.doi.org/10.14393/ufu.te.2018.7

Reciprocating hermetic compressors have been studied in several research programs aiming to noise reduction and understanding. The analysis of independent components is an important tool to allow identification of noise sources even without better information about their origin or how they had been mixed. The first stage of the present study is the tool understanding using known signal in order to figure out how the process is performed and its characteristics. Among the analyzed techniques, it was possible to observe that impulsive signals had been better recovered by time-domain techniques, while the harmonic signals had been better recovered by frequency-domain techniques. In addition, some algorithm stood out than other ones and extra noise source have improved performance results. When applying those mentioned techniques on the compressors, a limitation has been noticed in lower frequencies, because there is no phase difference between the analyzed signals and there is some correlation in signals syncing. Two components can be clearly identified, which are unbalanced forces at 60 Hz and gas pulsation for frequencies close to 500 Hz. The system has been tested on three different pressure levels: without any charge, 20 psi, and 60 psi. It has been found out that the gas pulsation can be better identified for the highest pressure level , while there is less correlation between the component and the signal for the lowest pressure level. Finally, the power pressure level contributions of recovered sources are estimated by a finite element method.