A probabilistic graphical model foundation for enabling predictive digital twins at scale

A unifying mathematical formulation is needed to move from one-off digital twins built through custom implementations robust twin at scale. This work proposes a probabilistic graphical model as formal representation of and its associated physical asset. We create an abstraction the asset–twin system set coupled dynamical systems, evolving over time their respective state spaces interacting via observed data control inputs. The definition this enables us draw upon well-established theory methods Bayesian statistics, systems theory. declarative general nature proposed make it rigorous yet flexible, enabling application scale in diverse range areas. demonstrate how instantiated enable structural unmanned aerial vehicle (UAV). calibrated using experimental UAV Its use dynamic decision-making then illustrated synthetic example where undergoes in-flight damage event dynamically updated sensor data. foundation ensures that calibration updating process principled, unified able entire fleet twins. for twins, demonstrates supports principled assimilation, optimal end-to-end uncertainty quantification.