Speaker
Description
Reliable magnet alignment requires assessing measurement capability before field implementation and selecting effective corrective actions when a measurement plan is insufficient. This study presents a task-specific, intervention-aware virtual metrology framework for the HALF GD04 magnet alignment. Laser tracker observation uncertainty is propagated through target coordinates, rigid-body pose estimation, and magnetic-center determination to eight transverse relative-position tasks. Four independent single-station datasets show that the predicted station-dependent uncertainty is consistent with observed inter-station differences. Counterfactual assessment indicates that longitudinal tracker relocation provides only limited and task-dependent improvement, with no insufficient task restored within the required threshold. Two-face measurement is therefore investigated as a protocol intervention. Paired AT960 observations reveal clear face-sensitive effects, although these effects cannot be reliably predicted for new targets from measurement geometry alone. Empirical propagation further shows that their task-level impact depends strongly on cross-target dependence, which is not identifiable from the current acquisition sequence. The framework enables pre-measurement comparison of engineering actions while explicitly reporting task trade-offs and evidence limitations. Future tunnel experiments will validate the proposed interventions using coordinated multi-target measurement cycles.