Figure 3.
Dynamic knowledge-graph-based approach toward automated closed-loop optimization. The real world layer demonstrates the existing physical entities, adapting from the experimentation setup of Jeraal et al.42 The dynamic knowledge graph layer hosts all the data generated during the experimentation and a digital twin of the experimentation apparatus. This layer is dynamic as it reflects and influences the status of the real world in real time. This synchronization is enforced by the agents in the active agents layer, which are instantiated from their ontological representation in the knowledge graph.