Fuzzy scheduling of a flexible assembly line through an evolutionary algorithm

In final assembly line operations, mixed-model lines allow one to reach different objectives, such as minimization of the line stop time (a productivity goal) and the component fluctuation (a JIT goal). This paper deals with product sequencing in mixed-model assembly lines, approaching the problem from a unique point of view: the minimization of the line stop time. Usually, such an analysis is performed through an approximate procedure because the problem data (i.e. the processing times) are estimated as deterministic values, but in real production this is a strong simplification; in fact, very often in a production environment, the data are vague, imprecise or uncertain. Then, the input data can be only estimated with a certain amount of uncertainty. When such uncertainty is primarily due to vagueness, it can usefully be formalized by using fuzzy mathematical tools. Such an approach in uncertainty modeling requires new methods for dealing with scheduling problems when the data are fuzzy. This paper proposes a new methodology for fuzzy scheduling in mixed-model assembly lines. Moreover, the object of the research is not only concentrated on fuzzy scheduling problem formalization but also on its optimization through proper powerful heuristic search tool developments, such as genetic algorithms and simulated annealing.