Lightweight CNN-Based Anomaly Detection for High Voltage Converter Modulators in the Spallation Neutron Source
arXiv:2605.31259v1 Announce Type: new Abstract: Unscheduled trips of high-power pulsed converters are a leading source of downtime at large accelerator facilities. At the Spallation Neutron Source (SNS), the High Voltage Converter Modulators (HVCMs) are consistently the second-largest contributor to lost beam time. Each HVCM pulse is recorded across sensor channels spanning currents, voltages, and magnetic fluxes, whose mutual interactions encode the operating state of the system. Fault precursors do not manifest uniformly across these channels: depending on fault type, they may alter the temp
The increasing complexity and demands on critical infrastructure, such as accelerator facilities, are driving the need for more sophisticated and efficient anomaly detection systems using AI.
This development streamlines the maintenance and improves the reliability of high-power scientific instruments by reducing unscheduled downtime through proactive fault detection.
The application of lightweight CNNs allows for real-time, resource-efficient anomaly detection in high-voltage systems, moving from reactive repairs to predictive maintenance.
- · Accelerator facilities
- · Scientific research institutions
- · AI/ML anomaly detection providers
- · Industrial control system manufacturers
- · Traditional anomaly detection methods
- · Manufacturers of less reliable high-power conversion equipment
Reduced operational costs and increased research throughput for facilities like the Spallation Neutron Source due to improved uptime.
Accelerated adoption of AI-driven predictive maintenance across other complex industrial and scientific infrastructure.
Potential for new standards in operational reliability for critical energy and research infrastructure, driven by AI's capabilities.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.LG