For objects (products, subsystems, components), digital twins as well as associated object agents are developed, which represent or elaborate and manage all quality-relevant attributes. Attributes are modeled individually for each object instance as Degree of Belief (DoB) distributions using Gaussian mixtures and Bayesian networks. An analogous information model will be developed for measurement and testing resources. New methods for the transformation of arbitrary information into DoB distributions are explored, so that an efficient Bayesian fusion and, based on instance-specific tolerance schemes, a productive sensor deployment planning becomes possible.

Principal Investigators