Adaptive Inventory Control

Abstract

Abstract The specific properties of adaptive control theory applications to inventory control (IC) problems are discussed. The variety of IC models is known. Along with rather simple linear dynamic (mainly multivariable) first order models the essentially nonlinear models are widely spread (IC systems with partially observed demand, obsolescence and diminution of stock, limited time of keeping, etc.). An important class of IC models is a class of demands-closed-loop models with inner stock-demand feedback. It is characteristic for IC problems to have both a priori and current information incomplete. The optimization criteria in the most of IC problems are piecewise linear or probabilistic ones. The property of acceleratedness is especially important for IC adaptive control algorithms, even at the expense of their asymptotic properties. The adaptive IC algorithms designed according to the mentioned properties of these problems are given in the paper, the simulation and practical application results are discussed.