SafeDivertor: AI-Based Heat Flux Reconstruction from Plasma Signals

Researchers have introduced SafeDivertor, a new framework for reconstructing divertor heat flux in magnetic-confinement fusion devices using macroscopic plasma state signals. The approach, detailed in a paper on arXiv (2608.05669), shifts from conventional infrared-based inversion methods that require post-discharge analysis and complex modeling of material properties, divertor geometry, and boundary conditions. Instead, SafeDivertor directly reconstructs time-resolved radial heat-flux profiles from multi-source signals available during discharge, enabling online monitoring. To support this, the team constructed DivMPS2HF, a multi-source discharge dataset that serves as a benchmark for signal-based reconstruction. The framework is task-driven and aims to improve plasma-wall interaction understanding and protect plasma-facing components. The paper was announced as a cross-type submission on arXiv, indicating potential interdisciplinary relevance. The work represents a significant step toward real-time heat flux analysis in fusion devices, potentially enhancing operational safety and efficiency.