Stochastic Scheduling of Manufacturing Systems

摘要

Fluid models can be used to approximate the behavior of a manufacturing plant which is modeled by a stochastic multi class queueing network, when the plant is in a transient state. We have recently shown that the behavior of the transient fluid model determines the stability of the stochastic queueing network. We have used these results to determine the global stability region (in which the stochastic network is stable under all work conserving policies) of two machine re-entrant lines. Fluid models can also be used to suggest heuristics for the control of a manufacturing plant in a transient state. We formulated the problem of draining a fluid re-entrant line with minimal average inventory and cycle time, as a continuous linear program and derived a special algorithm to solve it for the two machine case.