Transformation of semantic knowledge into simulation-based decision support

Abstract Simulation is capable to cope with the uncertain and dynamic nature of industrial value chains. However, in-depth system expertise is inevitable for mapping objects and constraints from the real world to a virtual model. This knowledge-intensity leads to long development times of respective projects, which contradicts the need for timely decision support. Since more and more companies use industrial knowledge graphs and ontologies to foster their knowledge management, this paper proposes a framework on how to efficiently derive a simulation model from such semantic knowledge bases. As part of the approach, a novel Simulation Ontology provides a standardized meta-model for hybrid simulations. Its instantiation enables the user to come up with a fully parameterized formal simulation model. Newly developed Mapping Rules facilitate this process by providing guidance on how to turn knowledge from existing ontologies, which describe the system to be simulated, into instances of the Simulation Ontology. The framework is completed by a parsing procedure for an automated transformation of this conceptual model into an executable one. This novel modeling approach makes model development more efficient by reducing its complexity. It is validated in a use case implementation from semiconductor manufacturing, where cross-domain knowledge was required in order to model and simulate the impacts of the COVID-19 pandemic on a global supply chain network.