Trust-Adaptive Multi-Diagnostic Weighting for Magnetically Confined Plasma State Estimation

Magnetically confined plasma experiments rely on heterogeneous diagnostic systems to estimate equilibrium and profile quantities in real time. Magnetic probes, interferometry, reflectometry, and soft X-ray arrays provide complementary but imperfect measurements of coupled plasma states. These diagnostics are subject to calibration drift, bandwidth limitations, and transient corruption during events such as edge-localized modes (ELMs), radiation spikes, and disruptions. Conventional reconstruction pipelines typically employ fixed weighting, static covariance tuning, or heuristic rejection thresholds. Such approaches can exhibit degraded robustness during fast transients or when individual diagnostics experience temporary faults. This work applies the Drift–Slew Fusion Bootstrap (DSFB) framework as a trust-adaptive weighting layer within multi-diagnostic plasma state estimation. DSFB operates by computing per-diagnostic residual envelopes and deriving continuous trust weights that attenuate the influence of channels exhibiting anomalous behavior. The underlying plasma model and reconstruction equations remain unchanged. We illustrate, via a synthetic equilibrium reconstruction scenario with injected ELM-like transient corruption, that adaptive trust weighting limits transient contamination within the reconstruction process. Analytical scaling arguments illustrate bounded influence under residual growth. The approach introduces minimal computational overhead and is compatible with existing real-time reconstruction pipelines. This study positions DSFB as an adaptive trust architecture for plasma diagnostics, enhancing transient robustness without altering established equilibrium solvers or physical modeling assumptions.