Feedforward Control in Plant Scenarios
Feedforward uses a known plan or disturbance to command actuators in advance, leaving feedback to correct only the residual error.
Anticipate, then correct
Pure feedback reacts only after an error appears, which is too slow when a large, predictable change is coming. Feedforward acts first: knowing the planned scenario or an incoming disturbance, it commands the actuator to the value that should produce the desired result, so the system is already close before feedback trims the remainder.
Scenario feedforward
A fusion pulse follows a planned scenario of current, heating, and fueling, so most actuator commands are computed in advance from that plan. Coil currents follow a scheduled waveform, heating ramps on a timetable, and cryogenic pre-cooling anticipates the pulse's heat load. Feedback then corrects for the difference between the model and reality, a much smaller and slower job.
Disturbance feedforward
When a disturbance is measurable before it fully acts, its effect can be canceled preemptively. For example, a scheduled current ramp adds a known heat load to the cryoplant, so the cooling controller can increase duty in advance rather than waiting for temperature to rise. This keeps regulated variables steadier than feedback alone could.
- Command actuators from the plan before error appears
- Scenario feedforward drives currents, heating, fueling
- Disturbance feedforward cancels known upsets early
- Feedback corrects only the residual
Combined structure
Real controllers combine feedforward and feedback: the feedforward carries the bulk of the command and the feedback handles model error, noise, and unmodeled disturbances. This split gives both a fast response and good steady accuracy. In the Kronos control architecture, described as a design and simulation model, scenario feedforward is central because fusion operation is highly scheduled, from magnet ramps to fueling.
Good feedforward makes feedback's job small, and a small feedback job is a stable one.