A Dynamic Bayesian Network Model for production and inventory control

In general, production quantities and delivered goods change randomly and consequently total stocks also change randomly. This paper deals with production and inventory control using a Dynamic Bayesian Network. The Bayesian Network is a probabilistic model which represents the qualitative dependence between two or more random variables by a graph structure, and the quantitative relations between individual variables by conditional probabilities. The probabilistic distribution of the total stock is calculated by propagation of probabilities on the network. Furthermore, a rule for adjustment of production quantities maintains the desired probabilities of exceeding the lower and upper limits on total stocks.