Speaker
Description
Beam faults at the HIAF (High Intensity Heavy-ion Accelerator Facility, Chian) iLinac rarely originate from a single subsystem; they emerge from coupled behaviours among the ion source, RFQ, superconducting cavities, LLRF, cryogenics, vacuum, magnets, and interlock systems. Conventional threshold alarms are subsystem-specific and often miss early precursors or produce false triggers. This oral presentation describes the design and initial deployment of a beam-based fault alarm and analysis platform. The platform ingests multi-source signals, maps each PV to its physical location along the linac, and applies data cleaning and spatial correlation. Rule-based and unsupervised anomaly detectors generate early warnings, while a fault knowledge base supports root-cause analysis. The platform coexists with hard interlocks: hardware protection remains the fast safety layer, while software provides early diagnosis and decision support. A closed loop from data acquisition to alarm, historical data retrieval, and algorithm invocation has been established and is being continuously refined. The platform aims to move HIAF operation from reactive firefighting toward predictive, data-driven maintenance.