IPAC2018 - List of Authors (Rainer, R.S.)

Paper	Title	Page
WEPAF022	Application of Machine Learning to Minimize Long Term Drifts in the NSLS-II Linac	1867
	R.P. Fliller, C. Gardner, P. Marino, R.S. Rainer, M. Santana, G.J. Weiner, X. Yang, E. Zeitler BNL, Upton, Long Island, New York, USA
	Funding: This manuscript has been authored by Brookhaven Science Associates, LLC under Contract No. DE-SC0012704 with the U.S. Department of Energy Machine Learning has proven itself as a useful technique in a variety of applications from image recognition to playing Go. Artificial Neural Networks have certain advantages when used as a feedforward system, such as the predicted correction relies on a model built from data. This allows for the Artificial Neural Network to compensate for effects that are difficult to model such as low level RF adjustments to compensate for long term drifts. The NSLS-II linac suffers from long terms drifts from a number of sources including thermal drifts and klystron gain variations. These drifts have an effect on the injection efficiency into the booster, and if left unchecked, portions of the bunch train may not be injected into the booster, and the storage ring bunch pattern will ultimately suffer. In this paper, we discuss the application of Artificial Neural Networks to compensate for long term drifts in the NSLS-II linear accelerator. The Artificial Neural Network is implemented in python allowing for rapid development of the network. We discuss the design and training of the network, along with results of using the network in operation.
DOI •	reference for this paper ※ https://doi.org/10.18429/JACoW-IPAC2018-WEPAF022
Export •	reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)

Paper

Title

Page

Application of Machine Learning to Minimize Long Term Drifts in the NSLS-II Linac

1867

R.P. Fliller, C. Gardner, P. Marino, R.S. Rainer, M. Santana, G.J. Weiner, X. Yang, E. Zeitler
BNL, Upton, Long Island, New York, USA

Funding: This manuscript has been authored by Brookhaven Science Associates, LLC under Contract No. DE-SC0012704 with the U.S. Department of Energy
Machine Learning has proven itself as a useful technique in a variety of applications from image recognition to playing Go. Artificial Neural Networks have certain advantages when used as a feedforward system, such as the predicted correction relies on a model built from data. This allows for the Artificial Neural Network to compensate for effects that are difficult to model such as low level RF adjustments to compensate for long term drifts. The NSLS-II linac suffers from long terms drifts from a number of sources including thermal drifts and klystron gain variations. These drifts have an effect on the injection efficiency into the booster, and if left unchecked, portions of the bunch train may not be injected into the booster, and the storage ring bunch pattern will ultimately suffer. In this paper, we discuss the application of Artificial Neural Networks to compensate for long term drifts in the NSLS-II linear accelerator. The Artificial Neural Network is implemented in python allowing for rapid development of the network. We discuss the design and training of the network, along with results of using the network in operation.

DOI •

reference for this paper ※ https://doi.org/10.18429/JACoW-IPAC2018-WEPAF022

Export •

reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)