Knowledge Graphs

The Digital Twin Graph is a property graph that encodes the physical and logical topology of an industrial facility. Nodes represent assets at multiple levels of granularity — from facility-level entities down to individual instruments, sensors, and firmware versions. Edges encode the coupling relationships that determine how events propagate between assets: physical couplings (piping, electrical interconnects, mechanical drive trains), control couplings (which PLCs control which actuators, which sensors feed which control loops), and dependency couplings (which safety systems protect which process units).

Agents use the Digital Twin Graph as their spatial and relational reasoning substrate. When an anomaly is detected on Compressor C-201, the agent traverses the graph to identify related assets: the intercooler connected to C-201's discharge, the lube oil system feeding C-201's bearings, the safety instrumented system protecting C-201 from overspeed, and the parallel compressor C-202 that could absorb C-201's load if C-201 is taken offline. This relational context is what enables the agent to reason about consequences and alternatives — capabilities that are impossible without a structured representation of asset topology.