پیش بینی و بازخورد دوره ای برای حفظ سطح موجودی هدف

In cases when a target inventory level is given and the demand is uncertain, it is difficult to provide an appropriate forecast and control policy that ensures a convergence to the inventory target. In this paper two different forecast formulae combined with two different feedback policies are investigated for a periodic review inventory system. A comparison based on analytical and simulation results is presented showing that the form of feedback is critical in case of large demand uncertainty.

Bonney (1994) and Bonney et al. (1994) surveyed papers and articles on inventory management. It seems that hat theories that had been studied could be classified into some groups by adopting certain criteria. Minimizing (average) inventory cost or maintaining (optimal) inventory level are such criteria. As an example of studies which are proposed to realize the minimal (average) inventory cost, theories of EOQ-Model are cited. EOQ-Model was first proposed by Arrow et al. (1951). By using this we can determine the ordering points of time and an ordering quantity. There is another group of studies which are proposed to maintain (optimal) inventory level. Ordering cycles are often determined by some rules of business practice. For such example we can cite the conditions which are presumed by the famous newsboy problem. Assume that a certain inventory level is given as the desired level. If a series of ordering points of time is given, maintaining the appropriate inventory level becomes the main problem. The pioneering study of Simmon (1952) dealt with the case in which a change in the inventory level is described by a differential equation. As an example of a study dealing difference equations, we can cite the work of Vassian (1955). These studies have attempted to apply the results of control theory to inventory control.

We assume that a series of points of time at which orders are placed is predetermined. A period between two successive ordering points of time is an ordering period. By expressing the change in inventory level during the period by a difference equation, the ordering quantity can be determined. The ordering quantity consists of two parts. One is the expected volume of demand, and the other is a feedback input. We proposed two forecast formulae. These two formulae for the volume of demand in the near future have been proposed under the assumption that there is trend in the change in demand. The two feedback policies were simultaneously planned. Therefore, we built four models so that we can examine the effect of introducing the feedback.