Extensão para controle on-line por atributo com erros de classificação: intervalo de inspeção variável, amostragem não-unitária, horizonte finito e infinito.

• Ano de defesa: 2009
• Autor(a) principal: Lupercio Franca Bessegato
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Minas Gerais; UFMG
• Palavras-chave em Português: Controle on-line por atributos; Cadeia de Markov; Planejamento econômico; Amostra não-unitária; Estatistica; Controle on-line de processos por atributos; Erros de classicação; Cadeias de Markov
• Link de acesso: http://hdl.handle.net/1843/BIRC-85CPT5

Taguchi et al. (1989) proposed an on-line process control for attribute procedure which consists in inspecting a single item at a xed sampling interval. Based on the result of each inspection, it is decided whether the process conforming fraction has changed or not. If an inspected item isdeclared non-conforming, it supposes that the production process is out-of-control and the process is adjusted to put it back to its initial condition. This paper studies economic policies that extend Taguchi's model paradigms. These proposals use the following: variable sampling interval, nonunitary sampling, variable sampling size and 3-level decisions. Finally, it studies a economic design for short-run production. It is considered that the classification system is imperfect. The averagecost expression for each proposed model is established in order to determine the optimal design parameter set. The decision process of the proposed models are illustrated by numerical examples. Keywords: on-line process control for attributes, economic design, misclassification errors, Markov chain, variable sampling interval, non-unitary sample, short-run production.