Condition Monitoring and Sensor Data
Predictive operations begin with dense, reliable sensing; the value of the models depends entirely on the quality of the data feeding them.
What is measured
A fusion plant carries thousands of sensors: magnetic pickup coils, thermocouples, strain gauges, pressure and flow meters, radiation monitors, and optical and spectroscopic diagnostics. Together they describe the state of the plasma, the magnets, the cooling loops, and the structure.
From raw signal to usable data
- Time synchronization so signals from different systems align to a common clock.
- Calibration so raw counts become physical units with known error.
- Quality flags so bad channels are detected and excluded automatically.
- Compression and storage so high-rate streams remain queryable later.
The data pipeline
Raw signals flow through a diagnostics pipeline that cleans, labels, and routes them to consumers: real-time control, the digital twin, and long-term archives. Latency budgets differ, so the pipeline splits a fast path for control from a thorough path for analysis.
Sensing what you cannot measure
Some quantities, like internal current profiles, cannot be measured directly. These are inferred by combining many sensors through state estimation. Good direct sensing makes those inferences trustworthy.
For Kronos machines
The Hyperion breeder and the tandem-mirror burner have different diagnostic sets, but both depend on the same principle: a model is only as good as the data behind it. Sensor placement is itself a design problem, optimized so the plant is observable.
Integrity
Sensor faults are treated as expected, not exceptional. The pipeline detects them, and downstream models are built to degrade gracefully when a channel drops, feeding anomaly detection.