Abstract: Particle accelerators are host to myriad nonlinear and complex physical phenomena. They often involve a multitude of interacting systems, are subject to tight performance demands, and should be able to run for extended periods of time with minimal interruptions. Often times, traditional control techniques cannot fully meet these requirements, and accelerator facilities rely heavily on highly skilled human “operators” to control the accelerator. This operational paradigm does not scale well outside of large facilities. Furthermore, the aforementioned control challenges become more acute in many new accelerator technologies and up-‐and-‐ coming applications. One promising avenue is to use machine learning and sophisticated control techniques inspired by artificial intelligence to address these challenges, particularly in light of recent theoretical and practical advances in these fields. Within machine learning and artificial intelligence, neural networks are useful for a wide variety of tasks. Neural networks are particularly well-‐suited to modeling, control, and diagnostic analysis of complex, nonlinear, and time-‐varying systems, as well as systems with large parameter spaces. Consequently, the use of neural network-‐based modeling and control techniques could be of significant benefit to particle accelerators. For the same reasons, particle accelerators are also ideal test-‐beds for these techniques. Many early attempts to apply neural networks to particle accelerators (mostly in the early 1990s) obtained mixed results due to the relative immaturity of the technology at that time. In this dissertation, we describe some of the challenges of particle accelerator control, highlight recent advances in neural network techniques in modeling and control, discuss some promising avenues for incorporating neural networks into particle accelerator control systems, and describe experimental studies in neural network control we are carrying out at Fermi National Accelerator Laboratory. The central aim is to advance the state of the art in particle accelerator control using neural network-‐based techniques through a combination of careful controller design and experimental testing at accelerator facilities. Such advances could increase the time-‐ and energy-‐efficiency of these machines, increase automation, improve beam quality, and further open the doors to existing and future applications of particle accelerators.
Adviser: Sandra Biedron Co-Adviser: Stephen Milton Non-ECE Member: Thomas Johnson, Environmental and Radiological Health Sciences Member 3: Edwin Chong, Electrical and Computer Engineering, Mathematics Addional Members: N/A
Publications: Edelen, A., et al. “Initial Experimental Results of a Machine Learning-Based Temperature Control System for an RF Gun.” Paper in proceedings of the 6th International Particle Accelerator Conference (IPAC), Richmond, VA, May 3-8, 2015.
Morin, A., et al. “Trajectory Response Studies at the Jefferson Laboratory Energy Recovery Linac and Free Electron Laser.” Paper in proceedings of the16th Annual Directed Energy Symposium, Huntsville, AL, March 10-14, 2014.
Program of Study: ECE680A3 ECE581A3 ECE641 ECE580B3 ECE656 ECE411 ECE520 ECE580A7