Skip Navigation

ECE Departmental Seminar 

Machine Learning Technologies in Particle Accelerators

Dr. Kevin Brown

Brookhaven National Laboratory

Friday, 5/4/18, 11:00am
Light Engineering 250

Abstract: Many techniques within Machine Learning (ML) are technologically mature enough to be incorporated into the operation of particle accelerators. Modern, large-scale accelerators are excellent places to employ these technologies. The interrelations between various machine subsystems are complicated and often nonlinear. The dynamics often involve large parameter spaces that evolve over multiple time scales. Many systems have behaviors that are difficult or impossible to model analytically. But automation in accelerator systems is approached with great caution. Accelerator components are capable of storing large amounts of energy (e.g., strings of superconducting magnets, superconducting RF systems, etc.). The particle beams can be very damaging, often with enough power to burn holes in vacuum chambers or damage magnet coils. ML technologies offer the promise of tackling these complex problems using relatively simple and stable methods.

In this talk I will introduce the different kinds of accelerators around the world where ML technologies are being employed, the kinds of problems they are helping to solve, and what problems we tackling at BNL using these technologies. 

Bio: Kevin Brown is a Physicist and the head of the Control Systems groups in the Collider Accelerator Department (C-AD) at Brookhaven National Laboratory, which operates the Relativistic Heavy Ion Collider (RHIC). Kevin is also leading the team that is designing and will build the control systems for the future electron-ion collider, eRHIC. Kevin is a member of the international executive committee overseeing the machine learning workshop series sponsored by the International Committee for Future Accelerators (ICFA), a member of the executive committee for the International Conference on Accelerator and Large Experimental Physics Control Systems (ICALEPCS), has been a member of the scientific and program committees for past conferences in that series and is chair/host for the upcoming ICALEPCS to be held in NYC in 2019. Currently, his research focus is on the use of Machine Learning technologies in Accelerator control systems as well as developing the infrastructure for control system simulations.