Robust Control of Perishable Inventory with Uncertain Lead Time Using Neural Networks and Genetic Algorithm

Abstract

The expansion of modern supply chains constantly triggers the need of maintaining resilience and agility for higher profit. There is a need to change the standard methods of inventory control to new approaches that are highly adaptable to uncertainties that emerged as a result of supply chains globalization. In this paper, a novel approach based on neural network, state-space control and robust optimization is proposed to support the perishable inventory replenishment decisions subject to uncertain lead times. We develop an approach based on the Wald criterion to compute optimal robust (i.e. “best of the worst” case) controller parameters. We incorporate lead-time specific perturbations through plausible scenarios using several lead times sets. Based on extensive numerical experiments, the obtained solutions highlight that the approach provides stable and robust solutions even for high lead times.