On the influence of state update interval length on the prediction success of decision support system in multi-site production environment

Abstract

Planning in a multi-site, non-mass production environment is a special challenge because of several sources of uncertainty. Unlike in mass production facilities, in our setting the current state at all sites cannot be determined easily and exactly due to the spatial distribution of sites and the low degree of automation. For re-planning in case of failures, the possible alternative actions have to be formalized on the decision making facility, where the possible alternatives will then be determined and evaluated. In this work, we will present the necessary components for an automated evaluation of alternatives and decision support procedure. The main challenges are the formalization of product plans including alternative steps and the non-automated collection or assessment of the distributed system state of all sites. In our experiments we evaluate different state update intervals and the effect on prediction accuracy. It turns out, that even sparse updates show significant improvement on the production time in comparison to only local static decisions.