Avaliação de um processo de eletrogalvanização por meio de modelagem estatística e cartas de controle

• Ano de defesa: 2015
• Autor(a) principal: Andara, Flávio Roberto
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• dARK ID: ark:/26339/001300000n1vp
• Idioma: por
• Instituição de defesa: Universidade Federal de Santa Maria; BR; Engenharia de Produção; UFSM; Programa de Pós-Graduação em Engenharia de Produção
• Palavras-chave em Português: Eletrogalvanização; Autocorrelação; Arima; Armax; MRLM; Electrolytic galvanizing; Auto-correlation; CNPQ::ENGENHARIAS::ENGENHARIA DE PRODUCAO
• Link de acesso: http://repositorio.ufsm.br/handle/1/8364

Quality tools, more specifically control charts, are important statistical resources to know and to monitor production processes. Their goal is to find the common and notable causes of a process to, through monitoring, increase the stability and, from it, assess if the process is under control. The dynamics of today s industrial activities has raised new requirements for good monitoring, and in that sense, new control tools have been developed and these are able to understand the new causal relationships among variables. The research shows the use of three modeling methodologies to treat autocorrelated data enabling to monitor a productive electroplating process. Initially, it was carried out a descriptive analysis for the verification of normality and independence and, afterwards, ARIMA from Box and Jenkins models, ARMAX models of multiple linear regression, MRLM, for the subsequent construction of waste control charts. In addition to the provided academic knowledge, it presents more than one application of control charts to the industrial environment, and also collaborates with the company where the research was developed showing which of the methods is more effective in controlling the production. The best result obtained by monitoring these three statistical methodologies work when confronted with the conventional control method, i.e., without treating the autocorrelation, it was used ARIMA model and a subsequent application of waste control charts derived from this modeling. The decision of the most effective methodology for modeling electroplating was defined by the number of points found out of the conventional limits established. The one that better captured the fluctuations of the process was obtained with the residues of ARIMA.