Utility-based Configuration of Learning Factories Using a Multidimensional, Multiple-choice Knapsack Problem

The paper presents a structural approach to configure the technical system of a learning factory by considering learning targets and maximizing the utility. Local scope conditions and intended competencies are used to operationalize requirements. The composition of the module-based technical system can be optimized by maximizing its overall utility. Therefore, an exact and efficient optimization algorithm is developed solving a multidimensional multiple-choice knapsack problem combined with a two-dimensional bin packing problem. Restrictions are the available budget and the useable area of the learning factory. As a result, the configured technical system enables optimal target orientation of the learning factory. This procedure is finally applied on the Process Learning Factory CiP.