Speaker
Description
RF and SRF operations require rapid decisions under tight time constraints while navigating machine data, logs, procedures and institutional knowledge. In many troubleshooting scenarios, the bottleneck is not diagnostic signal availability, but the time required to gather context, reconstruct event timelines, communicate status and transfer knowledge across shifts. This talk presents how AI can streamline RF/SRF operations while preserving engineering judgment. It focuses on document-grounded querying, multimodal interpretation of plots, screenshots and workflow acceleration for shift summaries, issue tracking and status reporting. Three use cases: faster trip triage, raw data processing and visualization and documentation, showing how AI can help convert scattered data into structured timelines, checklists and quick-reference guides. Responsible deployment is emphasized, with AI outputs treated as drafts and decision-support artifacts verified by personnel, measurements and procedures. Future work will describe a four-level LLM wiki framework for an RF/SRF “brain”: events/raw data acquisition, data cleaning/analysis/visualization, documentation and knowledge generation.
Funding Agency
Work supported by the U.S. Department of Energy, Office of Science under Cooperative Agreement DE-SC0023633, the State of Michigan, and Michigan State University.
| I have read and accept the Privacy Policy Statement | Yes |
|---|