Digital-Twin-Assisted Control
A digital twin runs a live model of the machine alongside the real one, supporting estimation, prediction, and validation of control actions.
What a digital twin adds
A digital twin is a model of the machine kept synchronized with the real device using live data. Unlike a flight simulator, which replaces the machine for testing, a twin runs concurrently with operation. It provides estimates of quantities that are not directly measured, predicts where the discharge is heading, and flags when reality diverges from the model.
Roles in control
- State estimation, inferring hidden variables from sparse measurements
- Short-horizon prediction, warning of an approaching limit
- What-if evaluation of a candidate control action before applying it
- Anomaly detection through model-versus-measurement residuals
Fidelity versus speed
A twin used inside or near the control loop must be fast, so it uses reduced models. A higher-fidelity twin can run between shots or in parallel for validation. Many programs keep a hierarchy: a fast twin for real-time use and a slow, accurate twin for analysis, with the fast one calibrated against the slow one.
Keeping it honest
A twin is only useful if it stays accurate. Its parameters are updated from operating data, and its predictions are continually checked against outcomes. A twin that drifts from the machine becomes worse than no twin, because it gives confident wrong answers, so validation is a permanent part of running one.
In the Kronos program
Kronos develops digital twins of the Hyperion breeder and the burner generators as part of the design effort, since the machines are in simulation ahead of construction beginning in the second quarter of 2027. The twins share reduced models with the flight simulator and the trajectory optimizer, so a scenario proven in one is consistent across all three. After first plasma, between-shot data will keep the twins calibrated.