Skip to main content
OSTI.GOVU.S. Department of Energy Office of Scientific and Technical Information
U.S. Department of Energy
Office of Scientific and Technical Information
 

FPGA Architectures for Distributed ML Systems for Real-time Beam Loss De-blending

Conference ·
OSTI ID:2006668
 [1];  [1];  [1];  [1];  [1];  [1];  [1];  [1];  [1];  [1];  [1];  [1];  [1];  [1];  [1];  [2];  [3];  [3];  [3];  [3] more »;  [3];  [3];  [3];  [3] « less
  1. Fermilab
  2. Fermilab; Northern Illinois U.
  3. Northwestern U.
+ Show Author Affiliations
The Accelerator Real-time Edge AI for Distributed Systems (READS) project’s goal is to create a Artificial Intelligence (AI) system for real-time beam loss de-blending within the accelerator enclosure, which houses two accelerators: the Main Injector (MI) and the Recycler Ring (RR). In periods of joint operation, when both machines contain high intensity beam, radioactive beam losses from MI and RR overlap on the enclosure’s beam loss monitoring Beam Loss Monitor (BLM) system, making it difficult to attribute those losses to a single machine. Incorrect diagnoses result in unnecessary downtime that incurs both financialand experimental cost. The ML system will automatically disentangle each machine’s contributions to those measured losses, while not disrupting the existing operations-critical functions of the BLM system. This paper will focus on the evolution of the architectures, which provided the high-frequency, low-latency collection of synchronized data streams to make real-time inferences. The ML models, used for learning both local and global machine signatures and producing high quality inferences based on raw BLM loss measurements, will only be discussed at a high-level.
Research Organization:
Fermi National Accelerator Laboratory (FNAL), Batavia, IL (United States)
Sponsoring Organization:
USDOE Office of Science (SC), High Energy Physics (HEP) (SC-25)
DOE Contract Number:
AC02-07CH11359
OSTI ID:
2006668
Report Number(s):
FERMILAB-CONF-23-281-AD; oai:inspirehep.net:2702980
Country of Publication:
United States
Language:
English

Similar Records

Real-Time Edge AI for Distributed Systems (READS): Progress on Beam Loss De-Blending for the Fermilab Main Injector and Recycler
Conference · Tue Aug 10 00:00:00 EDT 2021 · JACoW · OSTI ID:1833299
Semantic Regression for Disentangling Beam Losses in the Fermilab Main Injector and Recycler
Conference · Thu Oct 20 00:00:00 EDT 2022 · JACoW · OSTI ID:1909887
Disentangling Beam Losses in The Fermilab Main Injector Enclosure Using Real-Time Edge AI
Conference · Fri Dec 22 23:00:00 EST 2023 · JACoW · OSTI ID:2377381

Related Subjects

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS