Unleashing the Cognitive Digital Twin via Semantic Orchestration

The utility of embedding software representations of systems, equipment, and processes within monitoring and control software is not a new concept for solution developers. The authors have been incorporating cyber-physical “surrogate models”, a.k.a. Digital Twins (DTs), within operational intelligent systems software for several decades due to the many advantages such models provide. Because such systems are heterogenous and involve complex system-level interactions that depend on operational context, the addition of an ontologically derived semantic modeling layer on top of the DTs can greatly facilitate reasoning and improve situation awareness and decisioning. Such a semantic DT is referred to as a Cognitive DT (CDT), and the authors now present research and development of an Extended CDT (XDT) architecture that couples the CDT with a semantic orchestration engine (SOE). In the XDT, the semantic layer of the CDT forms an ontological bridge between the digital twin and the orchestration system software and supports the various forms of cognition necessary for evaluating and maximizing likelihoods related to mission success. The XDT can employ advanced data analytics for assessing the current and predicted future states of the system and its processes, evaluate any potential impacts to mission objectives, and then orchestrate any number of optimal responses. In addition to system components, the XDT includes model constructs for mission objectives, tasking, planning, and decisioning, situational workflows, and any number of concepts of operation (CONOPS). The semantic layer of the software also facilitates coding, code understanding, code debugging, testing, and especially system verification and validation (V & V). Furthermore, XDT software can collect semantic evidence and provide textual explanations for its decisions and conclusions. This ability for providing explanations has proven useful for V&V of XDT solution performance and opens the door for conversational run-time knowledge bases.