Singular perturbation methods in the synthesis of control policies for multi-stage production-inventory systems

In this paper, singular perturbation methods are used to synthesize control policies for a class of multi-stage production-inventory systems. It is shown that when the production-inventory sub-systems are connected in cascade, the control of each sub-system in isolation leads to undesirable transient behaviour. However, the control of the composite system is shown to yield improved transient behaviour. Furthermore, it is also shown that trade-off between rapid adjustments in production rate and large inventory-level deviation can be achieved by selecting an appropriate value for the feedback gain parameter.