Machine Learning System Improves Accelerator Diagnostics
A new machine learning system diagnoses particle accelerator component issues in real-time.
The Science
Particle accelerators are huge, complex machines. Scientists and engineers have designed and built a novel machine learning system to use with the Continuous Electron Beam Accelerator Facility (CEBAF). The system monitors structures called accelerator cavities inside the particle accelerator. These cavities impart energy to beams of electrons for exploring the nucleus of the atom. Problems in these cavities can cause the CEBAF to trip off like a fuse. In its first field test, the machine learning system correctly identified which of these cavities were tripping off about 85 percent of the time. About 78 percent of the time, the system also correctly identified what kind of fault caused each cavity to trip.
The Impact
CEBAF is the world’s primary research facility for exploring the nature of matter inside the atom’s nucleus. More than 1,650 nuclear physicists compete for limited research time to conduct experiments with CEBAF. Being able to identify potential problems in the machine early and quickly allows CEBAF operators to optimize the time available for experiments. The new machine learning system allows operators to identify the sources and types of problems nearly instantaneously. By quickly identifying problems, machine operators can resolve those problems faster. This reduces downtime and increases how much time experiments have on CEBAF and how much data they can collect. It also frees up time for machine experts to focus on other issues.
Summary
As every moment of research time is precious, staff at CEBAF, a Department of Energy Office of Science user facility, work to ensure that potential problems in CEBAF are found and quickly resolved. Previously, when an accelerator cavity or series of cavities faulted in CEBAF, accelerator experts needed to take time to diagnose the issue. Recently, researchers designed and built a machine learning system from scratch to aid in those efforts. The system was connected to the control system for about 20 percent of the accelerator cavities in the machine. In a two-week test of the system in March 2020, CEBAF experienced a few hundred faults that the system analyzed. It provided almost real-time feedback to operators, allowing them to use the information to recover faulted cavities quickly, reducing the time CEBAF’s electron beam was not available for research. Additional results are under analysis after further system tests with the machine learning system. If results are favorable, the team hopes to expand the system to more accelerator cavities.
Contact
Chris Tennant
Jefferson Lab
[email protected]
Funding
The foundation for this project was laid by work on a project led by Anna Shabalina, principal investigator on a proposal funded by Jefferson Lab’s Laboratory Directed Research and Development program. The project was later selected by DOE Nuclear Physics Program for a $1.35 million award to leverage machine learning to revolutionize experimentation and operations at user facilities in the coming years.
Publications
Tennant, C., et al., Superconducting radio-frequency cavity fault classification using machine learning at Jefferson Laboratory, Phys. Rev. Accel. Beams 23, 114601 (2020). [DOI: 10.1103/PhysRevAccelBeams.23.114601]Related Links
Machine Learning Improves Particle Accelerator Diagnostics, Jefferson Lab news release
DOE Funding Boosts Artificial Intelligence Research at Jefferson Lab, Jefferson Lab news release
Improving Particle Accelerators with Machine Learning, Jefferson Lab feature article
The Butterfly in the Machine, Jefferson Lab feature article
Department of Energy Announces $37 Million for Artificial Intelligence at Machine Learning at DOE Scientific User Facilities, DOE news release
Highlight Categories
Program: NP
Performer: DOE Laboratory , SC User Facilities , NP User Facilities , CEBAF
