ICALEPCS2021 - Classification: Feedback Control, Machine Tuning and Optimization

Paper	Title	Page
WEAL01	Image Processing Alignment Algorithms for the Optical Thomson Scattering Laser at the National Ignition Facility	528
	A.A.S. Awwal, T.S. Budge, R.R. Leach, R.R. Lowe-Webb, V.J. Miller Kamm, S. Patankar, B.P. Patel, K.C. Wilhelmsen LLNL, Livermore, California, USA
	Funding: This work performed under the auspices of the U.S. Department of Energy by Lawrence Livermore National Laboratory under Contract DE-AC52-07NA27344. Understanding plasma performance in the world’s largest and most energetic laser facility, the National Ignition Facility (NIF), is an important step to achieving the goal of inertial confinement fusion in a laboratory setting. The optical Thompson scattering (OTS) laser has been developed to understand the target implosion physics, especially for under-dense plasma conditions. A 5w probe beams can be set up for diagnosing various plasma densities. Just as the NIF laser with 192 laser beams are precisely aligned, the OTS system also requires precision alignment using a series of automated closed loop control steps. CCD images from the OTS laser (OTSL) beams are analyzed using a suite of image processing algorithm. The algorithms provide beam position measurements that are used to control motorized mirrors that steer beams to their defined desired location. In this paper, several alignment algorithms will be discussed with details on how they take advantage of various types of fiducials such as diffraction rings, contrasting squares and circles, octagons and very faint 5w laser beams. This is released as LLNL-ABS-821809
	Slides WEAL01 [1.303 MB]
DOI •	reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-WEAL01
About •	Received ※ 08 October 2021 Revised ※ 18 October 2021 Accepted ※ 21 November 2021 Issue date ※ 14 March 2022
Cite •	reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)

WEAL02	A Framework for High Level Machine Automation Based on Behavior Tree	534
	G. Gaio, P. Cinquegrana, S. Krecic, G. Scalamera, G. Strangolino, F. Tripaldi, M. Trovò, L. Zambon Elettra-Sincrotrone Trieste S.C.p.A., Basovizza, Italy
	In order to carry out complex tasks on particle accelerators, physicists and operators need to know the correct sequence of actions usually performed through a large number of graphical panels. The automation logics often embedded in the GUIs prevents its reuse by other programs, thus limiting the level of automation a control system can achieve. In order to overcome this limitation we have introduced a new automation framework for shifting the logics from GUIs to server side, where simple tasks can be easily organized, inspected and stacked up to build more complex actions. This tool is based on Behavior Trees (BT) which has been recently adopted in the gaming industry for in-game AI player opponents. They are able to create very complex tasks composed by simple decoupled self-contained tasks (nodes), regardless how they are implemented. The automation framework has been deployed in the Elettra and FERMI TANGO-based control systems to implement autonomous operations. A dedicated Qt GUI and a web interface allow to inspect the BTs and dynamically go through a tree, visualize the dependencies, monitor the execution and display any running action.
	Slides WEAL02 [1.809 MB]
DOI •	reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-WEAL02
About •	Received ※ 08 October 2021 Revised ※ 18 October 2021 Accepted ※ 21 November 2021 Issue date ※ 08 March 2022
Cite •	reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)

WEAL03	The Status of Fast Obit Feedback System of HEPS	540
	P. Zhu, Y.C. He, D.P. Jin, L. Zeng, Y.L. Zhang IHEP, Beijing, People’s Republic of China D.Y. Wang DNSC, Dongguan, People’s Republic of China L. Wang, X. Wu, Z.X. Xie, K. Xue IHEP CSNS, Guangdong Province, People’s Republic of China
	In order to further meet the needs of major national strategies and basic scientific research, High Energy Photon Source (HEPS) will be a high-performance fourth-generation synchrotron radiation source in Beijing, which will build more than 90 high-performance beamline stations. In order to ensure the high-performance operation of each beam line, the stability of the beam orbit near the light source output point is extremely important. As one of the key guarantees for the stability of the electron beam orbit, The FOFB system can suppress the beam orbit disturbance within a certain bandwidth to an acceptable range. This article introduces the currently progress of the FOFB system, including: the overall architecture scheme and key technical routes; the substation design following the ATCA mechanical architecture; the BPM data acquisition and high-speed transmission using high-performance Rocket I/O transmission Mechanism; embedded high-performance DSP for fast multiplication calculation to realize SVD, etc. The entire system design is progressing in an orderly manner.
	Slides WEAL03 [40.593 MB]
DOI •	reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-WEAL03
About •	Received ※ 19 October 2021 Revised ※ 22 October 2021 Accepted ※ 21 November 2021 Issue date ※ 23 February 2022
Cite •	reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)

WEPV001	Temperature Control for Beamline Precision Systems of Sirius/LNLS	607
	J.L. Brito Neto, R.R. Geraldes, F.R. Lena, M.A.L. Moraes, A.C. Piccino Neto, M. Saveri Silva, L.M. Volpe LNLS, Campinas, Brazil
	Funding: Ministry of Science, Technology and Innovation (MCTI) Precision beamline systems, such as monochromators and mirrors, as well as sample stages and sample holders, may require fine thermal management to meet performance targets. Regarding the optical elements, the main aspects of interest include substrate integrity, in case of high power loads and densities; wavefront preservation, due to thermal distortions of the optical surfaces; and beam stability, related to thermal drift. Concerning the sample, nanometer positioning control, for example, may be affected by thermal drifts and the power management of some electrical elements. This work presents the temperature control architecture developed in house for precision elements at the first beamlines of Sirius, the 4th-generation light source at the Brazilian Synchrotron Light Laboratory (LNLS). Taking some optical components as case studies, the predictive thermal-model-based approach, the system identification techniques, the controller design workflow and the implementation in hardware are described, as well as the temperature stability results.
	Poster WEPV001 [0.914 MB]
DOI •	reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-WEPV001
About •	Received ※ 15 October 2021 Accepted ※ 22 December 2021 Issue date ※ 21 February 2022
Cite •	reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)

WEPV002	Position Scanning Solutions at the TARUMÃ Station at the CARNAÚBA Beamline at Sirius/LNLS	613
	C.S.N.C. Bueno, L.G. Capovilla, R.R. Geraldes, L.C. Guedes, G.N. Kontogiorgos, L. Martins dos Santos, M.A.L. Moraes, G.B.Z.L. Moreno, A.C. Piccino Neto, J.R. Piton, H.C.N. Tolentino LNLS, Campinas, Brazil
	Funding: Ministry of Science, Technology and Innovation (MCTI) TARUMÃ is the sub-microprobe station of the CARNAÚBA beamline at Sirius/LNLS. Covering the range from 2.05 to 15keV, the probe consists of a fully-coherent monochromatic beam varying from 550 to 120nm with flux of up to 1e11ph/s/100mA after the achromatic focusing optics. Hence, positioning requirements span from nanometer-level errors for high-resolution experiments to fast continuous trajectories for high throughput, whereas a large flexibility is required for different sample setups and simultaneous multi-technique X-ray analyses, including tomography. To achieve this, the overall architecture of the station relies on a pragmatic sample positioning solution, with a rotation stage with a range of 220°, coarse stages for sub-micrometer resolution in a range of 20mm in XYZ and a fine piezo stage for nanometer resolution in a range of 0.3mm in XYZ. Typical scans consist of continuous raster 2D trajectories perpendicularly to the beam, over ranges that vary from tens to hundreds of micrometers, with acquisition times in range of milliseconds. Positioning is based on 4th order trajectories and feedforward, triggering includes the multiple detectors and data storage is addressed Geraldes, R.R., et al. ’Design and Commissioning of the TARUMÃ Station at the CARNAÚBA Beamline at Sirius/LNLS’ Proc. MEDSI20 (2020).
DOI •	reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-WEPV002
About •	Received ※ 10 October 2021 Accepted ※ 21 November 2021 Issue date ※ 05 February 2022
Cite •	reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)

WEPV003	The Dynamic Modeling and the Control Architecture of the New High-Dynamic Double-Crystal Monochromator (HD-DCM-Lite) for Sirius/LNLS	619
	G.S. de Albuquerque, J.L. Brito Neto, R.R. Geraldes, M.A.L. Moraes, A.V. Perna, M. Saveri Silva, M.S. Souza LNLS, Campinas, Brazil
	Funding: Ministry of Science, Technology and Innovation (MCTI) The High-Dynamic Double-Crystal Monochromator (HD-DCM) has been developed since 2015 at Sirius/LNLS with an innovative high-bandwidth mechatronic architecture to reach the unprecedented target of 10 nrad RMS (1 Hz - 2.5 kHz) in crystals parallelism also during energy flyscans. Now, for beamlines requiring a smaller energy range (3.1 to 43 keV, as compared to 2.3 to 72 keV), there is the opportunity to adapt the existing design towards the so-called HD-DCM-Lite. The control architecture of the HD-DCM is kept, reaching a 20 kHz control rate in NI’s CompactRIO (cRIO). Yet, the smaller gap stroke between crystals allows for removing the long-stroke mechanism and reducing the main inertia by a factor 6, not only simplifying the dynamics of the system, but also enabling faster energy scans. With sinusoidal scans of hundreds of eV up to 20 Hz, this creates an unparalleled bridge between slow step-scan DCMs, and channel-cut quick-EXAFS monochromators. This work presents the dynamic error budgeting and scanning perspectives for the HD-DCM-Lite, including feedback controller design options via loop shaping, feedforward considerations, and leader-follower control strategies.
	Poster WEPV003 [1.521 MB]
DOI •	reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-WEPV003
About •	Received ※ 13 October 2021 Accepted ※ 22 December 2021 Issue date ※ 26 December 2021
Cite •	reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)

WEPV005	Experiment Automation Using EPICS	625
	D.D. Cosic, M. Vićentijević RBI, Zagreb, Croatia
	Beam time at accelerator facilities around the world is very expensive and scarce, prompting the need for experiments to be performed as efficiently as possible. Efficiency of an accelerator facility is measured as a ratio of experiment time to beam optimization time. At RBI we have four ion sources, two accelerators, ten experimental end stations. We can obtain around 50 different ion species, each requiring a different set of parameters for optimal operation. Automating repetitive procedures can increase efficiency of an experiment and beam setup time. Currently, operators manually fine tunes the parameters to optimize the beam current. This process can be very long and requires many iterations. Automatic optimization of parameters can save valuable accelerator time. Based on a successful implementation of EPICS, the system was expanded to automate reoccurring procedures. To achieve this, a PLC was integrated into EPICS and our acquisition system was modified to communicate with devices through EPICS. This allowed us to use tools available in EPICS to do beam optimization much faster than a human operator can, and therefore significantly increased the efficiency of our facility.
	Poster WEPV005 [0.468 MB]
DOI •	reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-WEPV005
About •	Received ※ 08 October 2021 Accepted ※ 21 November 2021 Issue date ※ 16 February 2022
Cite •	reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)

WEPV006	Automated Operation of ITER Using Behavior Tree Semantics	628
	W. Van Herck, B. Bauvir, G. Ferro ITER Organization, St. Paul lez Durance, France
	The inherent complexity of the ITER machine and the diversity of the ways it will be operated in different phases, like commissioning or engineering operation, poses a great challenge for striking the right balance between operability, integration and automation. To facilitate the creation and execution of operational procedures in a robust and repeatable way, a software framework was developed: the Sequencer. As a supporting framework for tasks that are mostly goal-oriented, the Sequencer’s semantics are based on a behavior tree model that also supports concurrent flows of execution. In view of its intended use in very diverse situations, from small scale tests to full integrated operation, the architecture was designed to be composable and extensible from the start. User interactions with the Sequencer are fully decoupled and can be linked through dependency injection. The Sequencer library is currently feature-complete and comes with a command line interface for the encapsulation of procedures as system daemons or simple interactive use. It is highly maintainable due to its small and low complexity code base and dependencies to third party libraries are properly encapsulated.
DOI •	reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-WEPV006
About •	Received ※ 08 October 2021 Accepted ※ 21 November 2021 Issue date ※ 30 December 2021
Cite •	reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)

WEPV007	Machine Learning Projects at the 1.5-GeV Synchroton Light Source DELTA	631
	D. Schirmer, A. Althaus, S. Hüser, S. Khan, T. Schüngel DELTA, Dortmund, Germany
	In recent years, several machine learning (ML) based projects have been developed to support automated monitoring and operation of the DELTA electron storage ring facility. This includes self-regulating global and local orbit correction of the stored electron beam, betatron tune feedback as well as electron transfer rate (injection) optimization. Furthermore, the implementation for a ML-based chromaticity control is currently prepared. Some of these processes were initially simulated and then successfully transferred to real machine operation. This report provides an overview of the current status of these projects.
DOI •	reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-WEPV007
About •	Received ※ 10 October 2021 Accepted ※ 21 November 2021 Issue date ※ 02 February 2022
Cite •	reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)

WEPV008	Online Automatic Optimization of the Elettra Synchrotron	636
	G. Gaio, S. Krecic, F. Tripaldi Elettra-Sincrotrone Trieste S.C.p.A., Basovizza, Italy
	Online automatic optimization is a common practice in particle accelerators. Beside the tryouts based on Machine Learning, which are effective especially on non-linear systems and images but are very complex to tune and manage, one of the most simple and robust algorithms, the simplex Nelder Mead, is extensively used at Elettra to automatically optimize the synchrotron parameters. It is currently applied to optimize the efficiency of the booster injector by tuning the pre-injector energy, the trajectory and optics of the transfer lines, and the injection system of the storage ring. It has also been applied to maximize the intensity of the photon beam on a beamline by changing the electron beam position and angle inside the undulator. The optimization algorithm has been embedded in a TANGO device that also implements generic and configurable multi-input multi-output feedback systems. This optimization tool is usually included in a high level automation framework based on behavior trees in charge of the whole process of machine preparation for the experiments.
	Poster WEPV008 [1.600 MB]
DOI •	reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-WEPV008
About •	Received ※ 08 October 2021 Accepted ※ 26 January 2022 Issue date ※ 25 February 2022
Cite •	reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)

WEPV010	R&D of the KEK Linac Accelerator Tuning Using Machine Learning	640
	A. Hisano, M. Iwasaki OCU, Osaka, Japan H. Nagahara, Y. Nakashima, N. Takemura Osaka University, Institute for Datability Science, Oasaka, Japan T. Nakano RCNP, Osaka, Japan I. Satake, M. Satoh KEK, Ibaraki, Japan
	We have developed a machine-learning-based operation tuning scheme for the KEK e⁻/e⁺ injector linac (Linac), to improve the injection efficiency. The tuning scheme is based on the various accelerator operation data (control parameters, monitoring data and environmental data) of Linac. For the studies, we use the accumulated Linac operation data from 2018 to 2021. To solve the problems on the accelerator tuning of, 1. A lot of parameters (~1000) should be tuned, and these parameters are intricately correlated with each other; and 2. Continuous environmental change, due to temperature change, ground motion, tidal force, etc., affects to the operation tuning; We have developed, 1. Visualization of the accelerator parameters (~1000) trend/correlation distribution based on the dimensionality reduction using Variational Autoencoder (VAE), to see the long-term correlation between the accelerator operation parameters and the environmental data, and 2. Accelerator tuning method using the deep neural network, which is continuously updated with the short-term accelerator data to adapt the environment changes. In this presentation, we report the current status of the R&D.
	Poster WEPV010 [1.997 MB]
DOI •	reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-WEPV010
About •	Received ※ 10 October 2021 Revised ※ 19 October 2021 Accepted ※ 21 November 2021 Issue date ※ 11 January 2022
Cite •	reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)

WEPV011	Research on Correction of Beam Beta Function of HLS-II Storage Ring Based on Deep Learning	645
	Y.B. Yu, C. Li, W. Li, G. Liu, W. Xu, K. Xuan USTC/NSRL, Hefei, Anhui, People’s Republic of China
	The beam stability of the storage ring determines the light quality of synchrotron radiation. The beam stability of the storage ring will be affected by many factors ’such as magnetic field error, installation error, foundation vibration, temperature variation, etc., so it is inevitable to correct the beam optical parameters to improve the beam stability. In this paper, the deep learning technology is used to establish the HLS-II storage ring beam stability model, and the beam optical parameters can be corrected based on the model. The simulation results show that this method realizes the simulation correction of the Beta function of the HLS-II storage ring, and the correction accuracy precision meets the design requirements.
	Poster WEPV011 [2.142 MB]
DOI •	reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-WEPV011
About •	Received ※ 09 October 2021 Revised ※ 15 November 2021 Accepted ※ 17 November 2021 Issue date ※ 21 November 2021
Cite •	reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)

WEPV012	Beam Fast Recovery Study and Application for CAFe	648
	J.S. Li, Y.X. Chen, J. Wang, F. Yang, H. Zheng IMP/CAS, Lanzhou, People’s Republic of China
	Based on the MASAR (MAchine Snapshot, Archiving, and Retrieve) system, a beam fast recovery system was designed and tested in CAFe (Chinese ADS Front-end Demo Superconducting Linac) at IMP/CAS for high cur-rent CW (Continuous Wave) beam. The proton beam was accelerated to about 20 MeV with 23 SC (Superconduct-ing) cavities, and the maximum current reaches about 10 mA. The fast-recovery system plays a major role in the 100-hours-100-kW long-term test, during which the aver-age time of the beam recovery is 7 second, achieving the availability higher than 90%. The system verifies the possibility for high current beam fast recovery in CiADS (China initiative Accelerator Driven sub-critical System).
	Poster WEPV012 [0.469 MB]
DOI •	reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-WEPV012
About •	Received ※ 10 October 2021 Revised ※ 22 October 2021 Accepted ※ 21 November 2021 Issue date ※ 02 March 2022
Cite •	reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)

WEPV013	Design of Magnet Measurement System Based on Multi-Hall Sensor	653
	B.J. Wang, Y.H. Guo, R. Wang, N. Xie IMP/CAS, Lanzhou, People’s Republic of China
	High-precision magnetic field measurement and control technique significantly guarantees the accurate realization of the magnetic confinement of accelerators. Using real-time magnetic field intensity as the feedback to adjust the magnetic field current input can be a promising strategy. However, the measurement accuracy of the Hall-sensor is hard to meet feedback requirements because of multiple affection from external factors. Meanwhile, the NMR(Nuclear Magnetic Resonance sensor), which can provide high-precision magnetic field measurement, can hardly meet the requirements against the real-time control due to its strict requirements on the uniformity of the measured magnetic field, as well as its low data acquisition speed. Therefore, a magnetic field measurement system based on multi-Hall sensors is designed to solve this problem. Four Hall-sensors are used to measure the target magnetic field in this system. An Adaptive fusion algorithm is used to fused collected values to obtain the best estimate of the magnetic field intensity. This system effectively improves the accuracy of magnetic field measurement and ensures the instantaneity of the measurement.
	Poster WEPV013 [0.841 MB]
DOI •	reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-WEPV013
About •	Received ※ 09 October 2021 Revised ※ 22 October 2021 Accepted ※ 21 November 2021 Issue date ※ 06 December 2021
Cite •	reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)

WEPV015	Development of the RF Phase Scan Application for the Beam Energy Measurement at KOMAC	656
	S.Y. Cho, J.J. Dang, J.H. Kim, Y.G. Song KOMAC, KAERI, Gyeongju, Republic of Korea
	The Korea Multi-purpose Accelerator Complex (KOMAC) proton accelerator consists of 11 Drift Tube Linac (DTL) tanks, and each tank’s RF phase setting must be matched to increase synchronous acceleration of continuous tanks. A series of processes operate on the basis of JAVA and MatLAB languages, and the phase scanning program and the analytical program are classified and used independently. To integrate the two programs, the new integrated program of the RF scan application is developed based on python and epics scan module for the stability with some upgrade functions.
	Poster WEPV015 [1.051 MB]
DOI •	reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-WEPV015
About •	Received ※ 08 October 2021 Revised ※ 19 October 2021 Accepted ※ 21 November 2021 Issue date ※ 16 February 2022
Cite •	reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)

WEPV016	The Automatic LHC Collimator Beam-Based Alignment Software Package	659
	G. Azzopardi, B. Salvachua CERN, Geneva, Switzerland G. Valentino University of Malta, Information and Communication Technology, Msida, Malta
	The Large Hadron Collider (LHC) at CERN makes use of a complex collimation system to protect its sensitive equipment from unavoidable beam losses. The collimators are positioned around the beam respecting a strict transverse hierarchy. The position of each collimator is determined following a beam-based alignment technique which determines the required jaw settings for optimum performance. During the LHC Run 2 (2015-2018), a new automatic alignment software package was developed and used for collimator alignments throughout 2018. This paper discusses the usability and flexibility of this new package describing the implementation in detail, as well as the latest improvements and features in preparation for Run 3 starting in 2022. The automation has already successfully decreased the alignment time by 70% in 2018* and this paper explores how to further exploit this software package. Its implementation provides a solid foundation to automatically align any new collimation configurations in the future, as well as allows for further analysis and upgrade of its individual modules. G.Azzopardi, et al"Software Architecture for Automatic LHC Collimator Alignment using ML",ICALEPCS19. *G.Azzopardi, et al"Operational Results on the Fully-Automatic LHC Collimator Alignment",PRAB19.
	Poster WEPV016 [0.443 MB]
DOI •	reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-WEPV016
About •	Received ※ 07 October 2021 Revised ※ 22 October 2021 Accepted ※ 22 December 2021 Issue date ※ 26 December 2021
Cite •	reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)

WEPV018	The Linac4 Source Autopilot	665
	M. Peryt, M. Hrabia, D. Noll, R. Scrivens CERN, Geneva, Switzerland
	The Linac4 source is a 2MHz, RF driven, H⁻ ion source, using caesium injection to enhance H⁻ production and lower the electron to H⁻ ratio. The source operates with 800µs long pulses at 1.2 second intervals. The stability of the beam intensity from the source requires adjustment of parameters like RF power used for plasma heating. The Linac4 Source Autopilot improves the stability and uptime of the source, by using high-level automation to monitor and control Device parameters of the source, in a time range of minutes to days. This paper describes the Autopilot framework, which incorporates standard CERN accelerator Controls infrastructure, and enables users to add domain specific code for their needs. User code typically runs continuously, adapting Device settings based on acquisitions. Typical use cases are slow feedback systems and procedure automation (e.g. resetting equipment). The novelty of the Autopilot is the successful integration of the Controls software based predominantly on Java technologies, with domain specific user code written in Python. This allows users to leverage a robust Controls infrastructure, with minimal effort, using the agility of the Python ecosystem.
	Poster WEPV018 [4.371 MB]
DOI •	reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-WEPV018
About •	Received ※ 10 October 2021 Revised ※ 19 October 2021 Accepted ※ 22 December 2021 Issue date ※ 31 December 2021
Cite •	reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)

WEPV019	Renovation of the Beam-Based Feedback Controller in the LHC	671
	L. Grech, D. Alves, A. Calia, M. Hostettler, S. Jackson, J. Wenninger CERN, Meyrin, Switzerland G. Valentino University of Malta, Information and Communication Technology, Msida, Malta
	This work presents an extensive overview of the design choices and implementation of the Beam-Based Feedback System (BBFS) used in operation until the LHC Run 2. The main limitations of the BBFS are listed and a new design called BFCLHC, which uses the CERN Front-End Software Architecture (FESA), framework is proposed. The main implementation details and new features which improve upon the usability of the new design are then emphasised. Finally, a hardware agnostic testing framework developed by the LHC operations section is introduced.
DOI •	reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-WEPV019
About •	Received ※ 10 October 2021 Accepted ※ 21 November 2021 Issue date ※ 12 March 2022
Cite •	reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)

WEPV020	Learning to Lase: Machine Learning Prediction of FEL Beam Properties	677
	A.E. Pollard, D.J. Dunning STFC/DL/ASTeC, Daresbury, Warrington, Cheshire, United Kingdom M. Maheshwari STFC/DL, Daresbury, Warrington, Cheshire, United Kingdom
	Accurate prediction of longitudinal phase space and other properties of the electron beam are computationally expensive. In addition, some diagnostics are destructive in nature and/or cannot be readily accessed. Machine learning based virtual diagnostics can allow for the real-time generation of longitudinal phase space and other graphs, allowing for rapid parameter searches, and enabling operators to predict otherwise unavailable beam properties. We present a machine learning model for predicting a range of diagnostic screens along the accelerator beamline of a free-electron laser facility, conditional on linac and other parameters. Our model is a combination of a conditional variational autoencoder and a generative adversarial network, which generates high fidelity images that accurately match simulation data. Work to date is based on start-to-end simulation data, as a prototype for experimental applications.
	Poster WEPV020 [1.330 MB]
DOI •	reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-WEPV020
About •	Received ※ 10 October 2021 Revised ※ 22 October 2021 Accepted ※ 28 December 2021 Issue date ※ 25 February 2022
Cite •	reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)

WEPV021	Machine Learning for RF Breakdown Detection at CLARA	681
	A.E. Pollard, D.J. Dunning, A.J. Gilfellon STFC/DL/ASTeC, Daresbury, Warrington, Cheshire, United Kingdom
	Maximising the accelerating gradient of RF structures is fundamental to improving accelerator facility performance and cost-effectiveness. Structures must be subjected to a conditioning process before operational use, in which the gradient is gradually increased up to the operating value. A limiting effect during this process is breakdown or vacuum arcing, which can cause damage that limits the ultimate operating gradient. Techniques to efficiently condition the cavities while minimising the number of breakdowns are therefore important. In this paper, machine learning techniques are applied to detect breakdown events in RF pulse traces by approaching the problem as anomaly detection, using a variational autoencoder. This process detects deviations from normal operation and classifies them with near perfect accuracy. Offline data from various sources has been used to develop the techniques, which we aim to test at the CLARA facility at Daresbury Laboratory. These techniques could then be applied generally.
	Poster WEPV021 [1.565 MB]
DOI •	reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-WEPV021
About •	Received ※ 09 October 2021 Accepted ※ 21 November 2021 Issue date ※ 24 November 2021
Cite •	reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)

WEPV022	Sample Alignment in Neutron Scattering Experiments Using Deep Neural Network	686
	J.P. Edelen, K. Bruhwiler, A. Diaw, C.C. Hall RadiaSoft LLC, Boulder, Colorado, USA S. Calder ORNL RAD, Oak Ridge, Tennessee, USA C.M. Hoffmann ORNL, Oak Ridge, Tennessee, USA
	Funding: DOE Office of Science Office of Basic Energy Science SBIR award number DE-SC0021555 Access to neutron scattering centers, such as Oak Ridge National Laboratory (ORNL) and the NIST Center for Neutron Research, has provided beam energies to investigating a wide variety of applications such as particle physics, material science, and biology. In these experiments, the quality of collected data is very sensitive to sample and beam alignment, and stabilization of the experimental environment, requiring human intervention to tune the beam. While this procedure works, it is inefficient and time-consuming. In the work we present progress towards using machine learning to automate the alignment of a beamline in neutron scattering experiments. Our algorithm uses convolutional neural network to both learn a surrogate of the image data of the sample and to predict the sample contour using a u-net. We tested our algorithm on neutron camera images from the H2-BA powder diffractometer and the Topaz single crystal diffractometer beamlines of ORNL.
	Poster WEPV022 [4.472 MB]
DOI •	reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-WEPV022
About •	Received ※ 10 October 2021 Revised ※ 22 October 2021 Accepted ※ 21 December 2021 Issue date ※ 06 February 2022
Cite •	reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)

WEPV023	Development of a Smart Alarm System for the CEBAF Injector	691
	D.T. Abell, J.P. Edelen RadiaSoft LLC, Boulder, Colorado, USA B.G. Freeman, R. Kazimi, D.G. Moser, C. Tennant JLab, Newport News, Virginia, USA
	Funding: This material is based upon work supported by the U.S. Department of Energy, Office of Science, Office of Nuclear Physics under Award Number DE-SC0019682. RadiaSoft and Jefferson Laboratory are working together to develop a machine-learning-based smart alarm system for the CEBAF injector. Because of the injector’s large number of parameters and possible fault scenarios, it is highly desirable to have an autonomous alarm system that can quickly identify and diagnose unusual machine states. We present our work on artificial neural networks designed to identify such undesirable machine states. In particular, we test both auto-encoders and inverse models as possible tools for differentiating between normal and abnormal states. These models are being developed using both supervised and unsupervised learning techniques, and are being trained using CEBAF injector data collected during dedicated machine studies as well as during regular operations. Lastly, we discuss tradeoffs between the two types of models.
DOI •	reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-WEPV023
About •	Received ※ 10 October 2021 Accepted ※ 19 January 2022 Issue date ※ 14 March 2022
Cite •	reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)

WEPV024	X-Ray Beamline Control with Machine Learning and an Online Model	695
	B. Nash, D.T. Abell, D.L. Bruhwiler, E.G. Carlin, J.P. Edelen, M.V. Keilman, P. Moeller, R. Nagler, I.V. Pogorelov, S.D. Webb RadiaSoft LLC, Boulder, Colorado, USA Y. Du, A. Giles, J. Lynch, J. Maldonado, M.S. Rakitin, A. Walter BNL, Upton, New York, USA
	Funding: This material is based upon work supported by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences under contract DE-SC0020593. We present recent developments on control of x-ray beamlines for synchrotron light sources. Effective models of the x-ray transport are updated based on diagnostics data, and take the form of simplified physics models as well as learned models from scanning over mirror and slit configurations. We are developing this approach to beamline control in collaboration with several beamlines at the NSLS-II. By connecting our online models to the Blue-Sky framework, we enable a convenient interface between the operating machine and the model that may be applied to beamlines at multiple facilities involved in this collaborative software development.
DOI •	reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-WEPV024
About •	Received ※ 10 October 2021 Accepted ※ 21 November 2021 Issue date ※ 17 December 2021
Cite •	reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)

WEPV025	Initial Studies of Cavity Fault Prediction at Jefferson Laboratory	700
	L.S. Vidyaratne, A. Carpenter, R. Suleiman, C. Tennant, D.L. Turner JLab, Newport News, Virginia, USA K.M. Iftekharuddin, M. Rahman ODU, Norfolk, Virginia, USA
	Funding: This work supported by the U.S. Department of Energy, Office of Science, Office of Nuclear Physics under Contract No. DE-AC05-06OR23177. The Continuous Electron Beam Accelerator Facility (CEBAF) at Jefferson Laboratory is a CW recirculating linac that utilizes over 400 superconducting radio-frequency (SRF) cavities to accelerate electrons up to 12 GeV through 5-passes. Recent work has shown that, given RF signals from a cavity during a fault as input, machine learning approaches can accurately classify the fault type. In this paper we report on initial results of predicting a fault onset using only data prior to the failure event. A data set was constructed using time-series data immediately before a fault (’unstable’) and 1.5 seconds prior to a fault (’stable’) gathered from over 5,000 saved fault events. The data was used to train a binary classifier. The results gave key insights into the behavior of several fault types and provided motivation to investigate whether data prior to a failure event could also predict the type of fault. We discuss our method using a sliding window approach and report on initial results. Recent modifications to the low-level RF control system will provide access to streaming signals and we outline a path forward for leveraging deep learning on streaming data
	Poster WEPV025 [1.111 MB]
DOI •	reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-WEPV025
About •	Received ※ 08 October 2021 Revised ※ 19 October 2021 Accepted ※ 11 February 2022 Issue date ※ 05 March 2022
Cite •	reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)

THAL01	Machine Learning Tools Improve BESSY II Operation	784
	L. Vera Ramiréz, T. Birke, G. Hartmann, R. Müller, M. Ries, A. Schälicke, P. Schnizer HZB, Berlin, Germany
	At the HZB user facility BESSY II Machine Learning (ML) technologies aim at advanced analysis, automation, explainability and performance improvements for accelerator and beamline operation. The development of these tools is intertwined with improvements of the prediction part of the digital twin instances at BESSY II [] and the integration into the Bluesky Suite [,*]. On the accelerator side, several use cases have recently been identified, pipelines designed and models tested. Previous studies applied Deep Reinforcement Learning (RL) to booster current and injection efficiency. RL now tackles a more demanding scenario: the mitigation of harmonic orbit perturbations induced by external civil noise sources. This paper presents methodology, design and simulation phases as well as challenges and first results. Further ML use cases under study are, among others, anomaly detection prototypes with anomaly scores for individual features. [] P. Schnizer et. al, IPAC21 [] D. Allan, T. Caswell, S. Campbell and M. Rakitin, Synchrot. Radiat. News 32 19-22, 2019 [*] W. Smith et. al, this conference
	Slides THAL01 [9.849 MB]
DOI •	reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-THAL01
About •	Received ※ 08 October 2021 Revised ※ 24 October 2021 Accepted ※ 21 November 2021 Issue date ※ 29 January 2022
Cite •	reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)

THAL02	Bayesian Techniques for Accelerator Characterization and Control	791
	R.J. Roussel, A.L. Edelen, C.E. Mayes SLAC, Menlo Park, California, USA J.P. Gonzalez-Aguilera, Y.K. Kim University of Chicago, Chicago, Illinois, USA
	Funding: This work was supported by the U.S. National Science Foundation under Award No. PHY-1549132, the Center for Bright Beams. Accelerators and other large experimental facilities are complex, noisy systems that are difficult to characterize and control efficiently. Bayesian statistical modeling techniques are well suited to this task, as they minimize the number of experimental measurements needed to create robust models, by incorporating prior, but not necessarily exact, information about the target system. Furthermore, these models inherently take into account noisy and/or uncertain measurements and can react to time-varying systems. Here we will describe several advanced methods for using these models in accelerator characterization and optimization. First, we describe a method for rapid, turn-key exploration of input parameter spaces using little-to-no prior information about the target system. Second, we highlight the use of Multi-Objective Bayesian optimization towards efficiently characterizing the experimental Pareto front of a system. Throughout, we describe how unknown constraints and parameter modification costs are incorporated into these algorithms.
	Slides THAL02 [4.453 MB]
DOI •	reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-THAL02
About •	Received ※ 10 October 2021 Revised ※ 10 November 2021 Accepted ※ 21 November 2021 Issue date ※ 26 December 2021
Cite •	reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)

THAL03	Machine Learning Based Middle-Layer for Autonomous Accelerator Operation and Control	797
	S. Pioli, B. Buonomo, D. Di Giovenale, C. Di Giulio, L.G. Foggetta, G. Piermarini LNF-INFN, Frascati, Italy F. Cardelli, P. Ciuffetti INFN/LNF, Frascati, Italy V. Martinelli INFN/LNL, Legnaro (PD), Italy
	The Singularity project, led by National Laboratories of Frascati of the National Institute for Nuclear Physics (INFN-LNF), aim to develop automated machine-independent middle-layer to control accelerator operation through machine learning (ML) algorithms like Reinforcement Learning (RL) and Cluster integrated with accelerator’s sub-systems. In this work we will present architecture and of the middle-layer made with main purpose to drive user requests through the control framework backend and allow users to enjoy a better User Experience (UX) handling system performances without facing problems due to the interaction with control system. We will report the strategy to develop autonomous operation control with RL algorithms together with the fault detection capability improved by Clustering approach as breakdown and waveguide and RF cavity thermal stability monitor. Results of the first period of operation of this system, currently operating at the electron-positron LINAC of the Dafne complex in Frascati, autonomously controlling accelerator performance in terms of beam transport, beam current optimization and RF cavity phase-jitter compensation will be reported.
	Slides THAL03 [0.960 MB]
DOI •	reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-THAL03
About •	Received ※ 19 October 2021 Accepted ※ 22 December 2021 Issue date ※ 16 February 2022
Cite •	reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)

THAL04	Machine Learning Based Tuning and Diagnostics for the ATR Line at BNL	803
	J.P. Edelen, K. Bruhwiler, E.G. Carlin, C.C. Hall RadiaSoft LLC, Boulder, Colorado, USA K.A. Brown, V. Schoefer BNL, Upton, New York, USA
	Funding: This material is based upon work supported by the U.S. Department of Energy, Office of Science, Office of Nuclear Physics under Award Number DE-SC0019682. Over the past several years machine learning has increased in popularity for accelerator applications. We have been exploring the use of machine learning as a diagnostic and tuning tool for transfer line from the AGS to RHIC at Brookhaven National Laboratory. In our work, inverse models are used to either provide feed-forward corrections for beam steering or as a diagnostic to illuminate quadrupole magnets that have excitation errors. In this talk we present results on using machine learning for beam steering optimization for a range of different operating energies. We also demonstrate the use of inverse models for optical error diagnostics. Our results are from studies that use combine simulation and measurement data.
	Slides THAL04 [4.845 MB]
DOI •	reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-THAL04
About •	Received ※ 10 October 2021 Revised ※ 22 October 2021 Accepted ※ 06 February 2022 Issue date ※ 01 March 2022
Cite •	reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)

Paper

Title

Page

Image Processing Alignment Algorithms for the Optical Thomson Scattering Laser at the National Ignition Facility

528

A.A.S. Awwal, T.S. Budge, R.R. Leach, R.R. Lowe-Webb, V.J. Miller Kamm, S. Patankar, B.P. Patel, K.C. Wilhelmsen
LLNL, Livermore, California, USA

Funding: *This work performed under the auspices of the U.S. Department of Energy by Lawrence Livermore National Laboratory under Contract DE-AC52-07NA27344.
Understanding plasma performance in the world’s largest and most energetic laser facility, the National Ignition Facility (NIF), is an important step to achieving the goal of inertial confinement fusion in a laboratory setting. The optical Thompson scattering (OTS) laser has been developed to understand the target implosion physics, especially for under-dense plasma conditions. A 5w probe beams can be set up for diagnosing various plasma densities. Just as the NIF laser with 192 laser beams are precisely aligned, the OTS system also requires precision alignment using a series of automated closed loop control steps. CCD images from the OTS laser (OTSL) beams are analyzed using a suite of image processing algorithm. The algorithms provide beam position measurements that are used to control motorized mirrors that steer beams to their defined desired location. In this paper, several alignment algorithms will be discussed with details on how they take advantage of various types of fiducials such as diffraction rings, contrasting squares and circles, octagons and very faint 5w laser beams.
*This is released as LLNL-ABS-821809

Slides WEAL01 [1.303 MB]

DOI •

reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-WEAL01

About •

Received ※ 08 October 2021 Revised ※ 18 October 2021 Accepted ※ 21 November 2021 Issue date ※ 14 March 2022

Cite •

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

WEAL02

A Framework for High Level Machine Automation Based on Behavior Tree

534

G. Gaio, P. Cinquegrana, S. Krecic, G. Scalamera, G. Strangolino, F. Tripaldi, M. Trovò, L. Zambon
Elettra-Sincrotrone Trieste S.C.p.A., Basovizza, Italy

In order to carry out complex tasks on particle accelerators, physicists and operators need to know the correct sequence of actions usually performed through a large number of graphical panels. The automation logics often embedded in the GUIs prevents its reuse by other programs, thus limiting the level of automation a control system can achieve. In order to overcome this limitation we have introduced a new automation framework for shifting the logics from GUIs to server side, where simple tasks can be easily organized, inspected and stacked up to build more complex actions. This tool is based on Behavior Trees (BT) which has been recently adopted in the gaming industry for in-game AI player opponents. They are able to create very complex tasks composed by simple decoupled self-contained tasks (nodes), regardless how they are implemented. The automation framework has been deployed in the Elettra and FERMI TANGO-based control systems to implement autonomous operations. A dedicated Qt GUI and a web interface allow to inspect the BTs and dynamically go through a tree, visualize the dependencies, monitor the execution and display any running action.

Slides WEAL02 [1.809 MB]

DOI •

reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-WEAL02

About •

Received ※ 08 October 2021 Revised ※ 18 October 2021 Accepted ※ 21 November 2021 Issue date ※ 08 March 2022

Cite •

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

WEAL03

The Status of Fast Obit Feedback System of HEPS

540

P. Zhu, Y.C. He, D.P. Jin, L. Zeng, Y.L. Zhang
IHEP, Beijing, People’s Republic of China
D.Y. Wang
DNSC, Dongguan, People’s Republic of China
L. Wang, X. Wu, Z.X. Xie, K. Xue
IHEP CSNS, Guangdong Province, People’s Republic of China

In order to further meet the needs of major national strategies and basic scientific research, High Energy Photon Source (HEPS) will be a high-performance fourth-generation synchrotron radiation source in Beijing, which will build more than 90 high-performance beamline stations. In order to ensure the high-performance operation of each beam line, the stability of the beam orbit near the light source output point is extremely important. As one of the key guarantees for the stability of the electron beam orbit, The FOFB system can suppress the beam orbit disturbance within a certain bandwidth to an acceptable range. This article introduces the currently progress of the FOFB system, including: the overall architecture scheme and key technical routes; the substation design following the ATCA mechanical architecture; the BPM data acquisition and high-speed transmission using high-performance Rocket I/O transmission Mechanism; embedded high-performance DSP for fast multiplication calculation to realize SVD, etc. The entire system design is progressing in an orderly manner.

Slides WEAL03 [40.593 MB]

DOI •

reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-WEAL03

About •

Received ※ 19 October 2021 Revised ※ 22 October 2021 Accepted ※ 21 November 2021 Issue date ※ 23 February 2022

Cite •

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

WEPV001

Temperature Control for Beamline Precision Systems of Sirius/LNLS

607

J.L. Brito Neto, R.R. Geraldes, F.R. Lena, M.A.L. Moraes, A.C. Piccino Neto, M. Saveri Silva, L.M. Volpe
LNLS, Campinas, Brazil

Funding: Ministry of Science, Technology and Innovation (MCTI)
Precision beamline systems, such as monochromators and mirrors, as well as sample stages and sample holders, may require fine thermal management to meet performance targets. Regarding the optical elements, the main aspects of interest include substrate integrity, in case of high power loads and densities; wavefront preservation, due to thermal distortions of the optical surfaces; and beam stability, related to thermal drift. Concerning the sample, nanometer positioning control, for example, may be affected by thermal drifts and the power management of some electrical elements. This work presents the temperature control architecture developed in house for precision elements at the first beamlines of Sirius, the 4th-generation light source at the Brazilian Synchrotron Light Laboratory (LNLS). Taking some optical components as case studies, the predictive thermal-model-based approach, the system identification techniques, the controller design workflow and the implementation in hardware are described, as well as the temperature stability results.

Poster WEPV001 [0.914 MB]

DOI •

reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-WEPV001

About •

Received ※ 15 October 2021 Accepted ※ 22 December 2021 Issue date ※ 21 February 2022

Cite •

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

WEPV002

Position Scanning Solutions at the TARUMÃ Station at the CARNAÚBA Beamline at Sirius/LNLS

613

C.S.N.C. Bueno, L.G. Capovilla, R.R. Geraldes, L.C. Guedes, G.N. Kontogiorgos, L. Martins dos Santos, M.A.L. Moraes, G.B.Z.L. Moreno, A.C. Piccino Neto, J.R. Piton, H.C.N. Tolentino
LNLS, Campinas, Brazil

Funding: Ministry of Science, Technology and Innovation (MCTI)
TARUMÃ is the sub-microprobe station of the CARNAÚBA beamline at Sirius/LNLS*. Covering the range from 2.05 to 15keV, the probe consists of a fully-coherent monochromatic beam varying from 550 to 120nm with flux of up to 1e11ph/s/100mA after the achromatic focusing optics. Hence, positioning requirements span from nanometer-level errors for high-resolution experiments to fast continuous trajectories for high throughput, whereas a large flexibility is required for different sample setups and simultaneous multi-technique X-ray analyses, including tomography. To achieve this, the overall architecture of the station relies on a pragmatic sample positioning solution, with a rotation stage with a range of 220°, coarse stages for sub-micrometer resolution in a range of 20mm in XYZ and a fine piezo stage for nanometer resolution in a range of 0.3mm in XYZ. Typical scans consist of continuous raster 2D trajectories perpendicularly to the beam, over ranges that vary from tens to hundreds of micrometers, with acquisition times in range of milliseconds. Positioning is based on 4th order trajectories and feedforward, triggering includes the multiple detectors and data storage is addressed
* Geraldes, R.R., et al. ’Design and Commissioning of the TARUMÃ Station at the CARNAÚBA Beamline at Sirius/LNLS’ Proc. MEDSI20 (2020).

DOI •

reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-WEPV002

About •

Received ※ 10 October 2021 Accepted ※ 21 November 2021 Issue date ※ 05 February 2022

Cite •

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

WEPV003

The Dynamic Modeling and the Control Architecture of the New High-Dynamic Double-Crystal Monochromator (HD-DCM-Lite) for Sirius/LNLS

619

G.S. de Albuquerque, J.L. Brito Neto, R.R. Geraldes, M.A.L. Moraes, A.V. Perna, M. Saveri Silva, M.S. Souza
LNLS, Campinas, Brazil

Funding: Ministry of Science, Technology and Innovation (MCTI)
The High-Dynamic Double-Crystal Monochromator (HD-DCM) has been developed since 2015 at Sirius/LNLS with an innovative high-bandwidth mechatronic architecture to reach the unprecedented target of 10 nrad RMS (1 Hz - 2.5 kHz) in crystals parallelism also during energy flyscans. Now, for beamlines requiring a smaller energy range (3.1 to 43 keV, as compared to 2.3 to 72 keV), there is the opportunity to adapt the existing design towards the so-called HD-DCM-Lite. The control architecture of the HD-DCM is kept, reaching a 20 kHz control rate in NI’s CompactRIO (cRIO). Yet, the smaller gap stroke between crystals allows for removing the long-stroke mechanism and reducing the main inertia by a factor 6, not only simplifying the dynamics of the system, but also enabling faster energy scans. With sinusoidal scans of hundreds of eV up to 20 Hz, this creates an unparalleled bridge between slow step-scan DCMs, and channel-cut quick-EXAFS monochromators. This work presents the dynamic error budgeting and scanning perspectives for the HD-DCM-Lite, including feedback controller design options via loop shaping, feedforward considerations, and leader-follower control strategies.

Poster WEPV003 [1.521 MB]

DOI •

reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-WEPV003

About •

Received ※ 13 October 2021 Accepted ※ 22 December 2021 Issue date ※ 26 December 2021

Cite •

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

WEPV005

Experiment Automation Using EPICS

625

D.D. Cosic, M. Vićentijević
RBI, Zagreb, Croatia

Beam time at accelerator facilities around the world is very expensive and scarce, prompting the need for experiments to be performed as efficiently as possible. Efficiency of an accelerator facility is measured as a ratio of experiment time to beam optimization time. At RBI we have four ion sources, two accelerators, ten experimental end stations. We can obtain around 50 different ion species, each requiring a different set of parameters for optimal operation. Automating repetitive procedures can increase efficiency of an experiment and beam setup time. Currently, operators manually fine tunes the parameters to optimize the beam current. This process can be very long and requires many iterations. Automatic optimization of parameters can save valuable accelerator time. Based on a successful implementation of EPICS, the system was expanded to automate reoccurring procedures. To achieve this, a PLC was integrated into EPICS and our acquisition system was modified to communicate with devices through EPICS. This allowed us to use tools available in EPICS to do beam optimization much faster than a human operator can, and therefore significantly increased the efficiency of our facility.

Poster WEPV005 [0.468 MB]

DOI •

reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-WEPV005

About •

Received ※ 08 October 2021 Accepted ※ 21 November 2021 Issue date ※ 16 February 2022

Cite •

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

WEPV006

Automated Operation of ITER Using Behavior Tree Semantics

628

W. Van Herck, B. Bauvir, G. Ferro
ITER Organization, St. Paul lez Durance, France

The inherent complexity of the ITER machine and the diversity of the ways it will be operated in different phases, like commissioning or engineering operation, poses a great challenge for striking the right balance between operability, integration and automation. To facilitate the creation and execution of operational procedures in a robust and repeatable way, a software framework was developed: the Sequencer. As a supporting framework for tasks that are mostly goal-oriented, the Sequencer’s semantics are based on a behavior tree model that also supports concurrent flows of execution. In view of its intended use in very diverse situations, from small scale tests to full integrated operation, the architecture was designed to be composable and extensible from the start. User interactions with the Sequencer are fully decoupled and can be linked through dependency injection. The Sequencer library is currently feature-complete and comes with a command line interface for the encapsulation of procedures as system daemons or simple interactive use. It is highly maintainable due to its small and low complexity code base and dependencies to third party libraries are properly encapsulated.

DOI •

reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-WEPV006

About •

Received ※ 08 October 2021 Accepted ※ 21 November 2021 Issue date ※ 30 December 2021

Cite •

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

WEPV007

Machine Learning Projects at the 1.5-GeV Synchroton Light Source DELTA

631

D. Schirmer, A. Althaus, S. Hüser, S. Khan, T. Schüngel
DELTA, Dortmund, Germany

In recent years, several machine learning (ML) based projects have been developed to support automated monitoring and operation of the DELTA electron storage ring facility. This includes self-regulating global and local orbit correction of the stored electron beam, betatron tune feedback as well as electron transfer rate (injection) optimization. Furthermore, the implementation for a ML-based chromaticity control is currently prepared. Some of these processes were initially simulated and then successfully transferred to real machine operation. This report provides an overview of the current status of these projects.

DOI •

reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-WEPV007

About •

Received ※ 10 October 2021 Accepted ※ 21 November 2021 Issue date ※ 02 February 2022

Cite •

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

WEPV008

Online Automatic Optimization of the Elettra Synchrotron

636

G. Gaio, S. Krecic, F. Tripaldi
Elettra-Sincrotrone Trieste S.C.p.A., Basovizza, Italy

Online automatic optimization is a common practice in particle accelerators. Beside the tryouts based on Machine Learning, which are effective especially on non-linear systems and images but are very complex to tune and manage, one of the most simple and robust algorithms, the simplex Nelder Mead, is extensively used at Elettra to automatically optimize the synchrotron parameters. It is currently applied to optimize the efficiency of the booster injector by tuning the pre-injector energy, the trajectory and optics of the transfer lines, and the injection system of the storage ring. It has also been applied to maximize the intensity of the photon beam on a beamline by changing the electron beam position and angle inside the undulator. The optimization algorithm has been embedded in a TANGO device that also implements generic and configurable multi-input multi-output feedback systems. This optimization tool is usually included in a high level automation framework based on behavior trees in charge of the whole process of machine preparation for the experiments.

Poster WEPV008 [1.600 MB]

DOI •

reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-WEPV008

About •

Received ※ 08 October 2021 Accepted ※ 26 January 2022 Issue date ※ 25 February 2022

Cite •

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

WEPV010

R&D of the KEK Linac Accelerator Tuning Using Machine Learning

640

A. Hisano, M. Iwasaki
OCU, Osaka, Japan
H. Nagahara, Y. Nakashima, N. Takemura
Osaka University, Institute for Datability Science, Oasaka, Japan
T. Nakano
RCNP, Osaka, Japan
I. Satake, M. Satoh
KEK, Ibaraki, Japan

We have developed a machine-learning-based operation tuning scheme for the KEK e⁻/e⁺ injector linac (Linac), to improve the injection efficiency. The tuning scheme is based on the various accelerator operation data (control parameters, monitoring data and environmental data) of Linac. For the studies, we use the accumulated Linac operation data from 2018 to 2021. To solve the problems on the accelerator tuning of, 1. A lot of parameters (~1000) should be tuned, and these parameters are intricately correlated with each other; and 2. Continuous environmental change, due to temperature change, ground motion, tidal force, etc., affects to the operation tuning; We have developed, 1. Visualization of the accelerator parameters (~1000) trend/correlation distribution based on the dimensionality reduction using Variational Autoencoder (VAE), to see the long-term correlation between the accelerator operation parameters and the environmental data, and 2. Accelerator tuning method using the deep neural network, which is continuously updated with the short-term accelerator data to adapt the environment changes. In this presentation, we report the current status of the R&D.

Poster WEPV010 [1.997 MB]

DOI •

reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-WEPV010

About •

Received ※ 10 October 2021 Revised ※ 19 October 2021 Accepted ※ 21 November 2021 Issue date ※ 11 January 2022

Cite •

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

WEPV011

Research on Correction of Beam Beta Function of HLS-II Storage Ring Based on Deep Learning

645

Y.B. Yu, C. Li, W. Li, G. Liu, W. Xu, K. Xuan
USTC/NSRL, Hefei, Anhui, People’s Republic of China

The beam stability of the storage ring determines the light quality of synchrotron radiation. The beam stability of the storage ring will be affected by many factors ’such as magnetic field error, installation error, foundation vibration, temperature variation, etc., so it is inevitable to correct the beam optical parameters to improve the beam stability. In this paper, the deep learning technology is used to establish the HLS-II storage ring beam stability model, and the beam optical parameters can be corrected based on the model. The simulation results show that this method realizes the simulation correction of the Beta function of the HLS-II storage ring, and the correction accuracy precision meets the design requirements.

Poster WEPV011 [2.142 MB]

DOI •

reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-WEPV011

About •

Received ※ 09 October 2021 Revised ※ 15 November 2021 Accepted ※ 17 November 2021 Issue date ※ 21 November 2021

Cite •

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

WEPV012

Beam Fast Recovery Study and Application for CAFe

648

J.S. Li, Y.X. Chen, J. Wang, F. Yang, H. Zheng
IMP/CAS, Lanzhou, People’s Republic of China

Based on the MASAR (MAchine Snapshot, Archiving, and Retrieve) system, a beam fast recovery system was designed and tested in CAFe (Chinese ADS Front-end Demo Superconducting Linac) at IMP/CAS for high cur-rent CW (Continuous Wave) beam. The proton beam was accelerated to about 20 MeV with 23 SC (Superconduct-ing) cavities, and the maximum current reaches about 10 mA. The fast-recovery system plays a major role in the 100-hours-100-kW long-term test, during which the aver-age time of the beam recovery is 7 second, achieving the availability higher than 90%. The system verifies the possibility for high current beam fast recovery in CiADS (China initiative Accelerator Driven sub-critical System).

Poster WEPV012 [0.469 MB]

DOI •

reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-WEPV012

About •

Received ※ 10 October 2021 Revised ※ 22 October 2021 Accepted ※ 21 November 2021 Issue date ※ 02 March 2022

Cite •

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

WEPV013

Design of Magnet Measurement System Based on Multi-Hall Sensor

653

B.J. Wang, Y.H. Guo, R. Wang, N. Xie
IMP/CAS, Lanzhou, People’s Republic of China

High-precision magnetic field measurement and control technique significantly guarantees the accurate realization of the magnetic confinement of accelerators. Using real-time magnetic field intensity as the feedback to adjust the magnetic field current input can be a promising strategy. However, the measurement accuracy of the Hall-sensor is hard to meet feedback requirements because of multiple affection from external factors. Meanwhile, the NMR(Nuclear Magnetic Resonance sensor), which can provide high-precision magnetic field measurement, can hardly meet the requirements against the real-time control due to its strict requirements on the uniformity of the measured magnetic field, as well as its low data acquisition speed. Therefore, a magnetic field measurement system based on multi-Hall sensors is designed to solve this problem. Four Hall-sensors are used to measure the target magnetic field in this system. An Adaptive fusion algorithm is used to fused collected values to obtain the best estimate of the magnetic field intensity. This system effectively improves the accuracy of magnetic field measurement and ensures the instantaneity of the measurement.

Poster WEPV013 [0.841 MB]

DOI •

reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-WEPV013

About •

Received ※ 09 October 2021 Revised ※ 22 October 2021 Accepted ※ 21 November 2021 Issue date ※ 06 December 2021

Cite •

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

WEPV015

Development of the RF Phase Scan Application for the Beam Energy Measurement at KOMAC

656

S.Y. Cho, J.J. Dang, J.H. Kim, Y.G. Song
KOMAC, KAERI, Gyeongju, Republic of Korea

The Korea Multi-purpose Accelerator Complex (KOMAC) proton accelerator consists of 11 Drift Tube Linac (DTL) tanks, and each tank’s RF phase setting must be matched to increase synchronous acceleration of continuous tanks. A series of processes operate on the basis of JAVA and MatLAB languages, and the phase scanning program and the analytical program are classified and used independently. To integrate the two programs, the new integrated program of the RF scan application is developed based on python and epics scan module for the stability with some upgrade functions.

Poster WEPV015 [1.051 MB]

DOI •

reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-WEPV015

About •

Received ※ 08 October 2021 Revised ※ 19 October 2021 Accepted ※ 21 November 2021 Issue date ※ 16 February 2022

Cite •

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

WEPV016

The Automatic LHC Collimator Beam-Based Alignment Software Package

659

G. Azzopardi, B. Salvachua
CERN, Geneva, Switzerland
G. Valentino
University of Malta, Information and Communication Technology, Msida, Malta

The Large Hadron Collider (LHC) at CERN makes use of a complex collimation system to protect its sensitive equipment from unavoidable beam losses. The collimators are positioned around the beam respecting a strict transverse hierarchy. The position of each collimator is determined following a beam-based alignment technique which determines the required jaw settings for optimum performance. During the LHC Run 2 (2015-2018), a new automatic alignment software package was developed and used for collimator alignments throughout 2018*. This paper discusses the usability and flexibility of this new package describing the implementation in detail, as well as the latest improvements and features in preparation for Run 3 starting in 2022. The automation has already successfully decreased the alignment time by 70% in 2018** and this paper explores how to further exploit this software package. Its implementation provides a solid foundation to automatically align any new collimation configurations in the future, as well as allows for further analysis and upgrade of its individual modules.
*G.Azzopardi, et al"Software Architecture for Automatic LHC Collimator Alignment using ML",ICALEPCS19.
**G.Azzopardi, et al"Operational Results on the Fully-Automatic LHC Collimator Alignment",PRAB19.

Poster WEPV016 [0.443 MB]

DOI •

reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-WEPV016

About •

Received ※ 07 October 2021 Revised ※ 22 October 2021 Accepted ※ 22 December 2021 Issue date ※ 26 December 2021

Cite •

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

WEPV018

The Linac4 Source Autopilot

665

M. Peryt, M. Hrabia, D. Noll, R. Scrivens
CERN, Geneva, Switzerland

The Linac4 source is a 2MHz, RF driven, H⁻ ion source, using caesium injection to enhance H⁻ production and lower the electron to H⁻ ratio. The source operates with 800µs long pulses at 1.2 second intervals. The stability of the beam intensity from the source requires adjustment of parameters like RF power used for plasma heating. The Linac4 Source Autopilot improves the stability and uptime of the source, by using high-level automation to monitor and control Device parameters of the source, in a time range of minutes to days. This paper describes the Autopilot framework, which incorporates standard CERN accelerator Controls infrastructure, and enables users to add domain specific code for their needs. User code typically runs continuously, adapting Device settings based on acquisitions. Typical use cases are slow feedback systems and procedure automation (e.g. resetting equipment). The novelty of the Autopilot is the successful integration of the Controls software based predominantly on Java technologies, with domain specific user code written in Python. This allows users to leverage a robust Controls infrastructure, with minimal effort, using the agility of the Python ecosystem.

Poster WEPV018 [4.371 MB]

DOI •

reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-WEPV018

About •

Received ※ 10 October 2021 Revised ※ 19 October 2021 Accepted ※ 22 December 2021 Issue date ※ 31 December 2021

Cite •

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

WEPV019

Renovation of the Beam-Based Feedback Controller in the LHC

671

L. Grech, D. Alves, A. Calia, M. Hostettler, S. Jackson, J. Wenninger
CERN, Meyrin, Switzerland
G. Valentino
University of Malta, Information and Communication Technology, Msida, Malta

This work presents an extensive overview of the design choices and implementation of the Beam-Based Feedback System (BBFS) used in operation until the LHC Run 2. The main limitations of the BBFS are listed and a new design called BFCLHC, which uses the CERN Front-End Software Architecture (FESA), framework is proposed. The main implementation details and new features which improve upon the usability of the new design are then emphasised. Finally, a hardware agnostic testing framework developed by the LHC operations section is introduced.

DOI •

reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-WEPV019

About •

Received ※ 10 October 2021 Accepted ※ 21 November 2021 Issue date ※ 12 March 2022

Cite •

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

WEPV020

Learning to Lase: Machine Learning Prediction of FEL Beam Properties

677

A.E. Pollard, D.J. Dunning
STFC/DL/ASTeC, Daresbury, Warrington, Cheshire, United Kingdom
M. Maheshwari
STFC/DL, Daresbury, Warrington, Cheshire, United Kingdom

Accurate prediction of longitudinal phase space and other properties of the electron beam are computationally expensive. In addition, some diagnostics are destructive in nature and/or cannot be readily accessed. Machine learning based virtual diagnostics can allow for the real-time generation of longitudinal phase space and other graphs, allowing for rapid parameter searches, and enabling operators to predict otherwise unavailable beam properties. We present a machine learning model for predicting a range of diagnostic screens along the accelerator beamline of a free-electron laser facility, conditional on linac and other parameters. Our model is a combination of a conditional variational autoencoder and a generative adversarial network, which generates high fidelity images that accurately match simulation data. Work to date is based on start-to-end simulation data, as a prototype for experimental applications.

Poster WEPV020 [1.330 MB]

DOI •

reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-WEPV020

About •

Received ※ 10 October 2021 Revised ※ 22 October 2021 Accepted ※ 28 December 2021 Issue date ※ 25 February 2022

Cite •

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

WEPV021

Machine Learning for RF Breakdown Detection at CLARA

681

A.E. Pollard, D.J. Dunning, A.J. Gilfellon
STFC/DL/ASTeC, Daresbury, Warrington, Cheshire, United Kingdom

Maximising the accelerating gradient of RF structures is fundamental to improving accelerator facility performance and cost-effectiveness. Structures must be subjected to a conditioning process before operational use, in which the gradient is gradually increased up to the operating value. A limiting effect during this process is breakdown or vacuum arcing, which can cause damage that limits the ultimate operating gradient. Techniques to efficiently condition the cavities while minimising the number of breakdowns are therefore important. In this paper, machine learning techniques are applied to detect breakdown events in RF pulse traces by approaching the problem as anomaly detection, using a variational autoencoder. This process detects deviations from normal operation and classifies them with near perfect accuracy. Offline data from various sources has been used to develop the techniques, which we aim to test at the CLARA facility at Daresbury Laboratory. These techniques could then be applied generally.

Poster WEPV021 [1.565 MB]

DOI •

reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-WEPV021

About •

Received ※ 09 October 2021 Accepted ※ 21 November 2021 Issue date ※ 24 November 2021

Cite •

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

WEPV022

Sample Alignment in Neutron Scattering Experiments Using Deep Neural Network

686

J.P. Edelen, K. Bruhwiler, A. Diaw, C.C. Hall
RadiaSoft LLC, Boulder, Colorado, USA
S. Calder
ORNL RAD, Oak Ridge, Tennessee, USA
C.M. Hoffmann
ORNL, Oak Ridge, Tennessee, USA

Funding: DOE Office of Science Office of Basic Energy Science SBIR award number DE-SC0021555
Access to neutron scattering centers, such as Oak Ridge National Laboratory (ORNL) and the NIST Center for Neutron Research, has provided beam energies to investigating a wide variety of applications such as particle physics, material science, and biology. In these experiments, the quality of collected data is very sensitive to sample and beam alignment, and stabilization of the experimental environment, requiring human intervention to tune the beam. While this procedure works, it is inefficient and time-consuming. In the work we present progress towards using machine learning to automate the alignment of a beamline in neutron scattering experiments. Our algorithm uses convolutional neural network to both learn a surrogate of the image data of the sample and to predict the sample contour using a u-net. We tested our algorithm on neutron camera images from the H2-BA powder diffractometer and the Topaz single crystal diffractometer beamlines of ORNL.

Poster WEPV022 [4.472 MB]

DOI •

reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-WEPV022

About •

Received ※ 10 October 2021 Revised ※ 22 October 2021 Accepted ※ 21 December 2021 Issue date ※ 06 February 2022

Cite •

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

WEPV023

Development of a Smart Alarm System for the CEBAF Injector

691

D.T. Abell, J.P. Edelen
RadiaSoft LLC, Boulder, Colorado, USA
B.G. Freeman, R. Kazimi, D.G. Moser, C. Tennant
JLab, Newport News, Virginia, USA

Funding: This material is based upon work supported by the U.S. Department of Energy, Office of Science, Office of Nuclear Physics under Award Number DE-SC0019682.
RadiaSoft and Jefferson Laboratory are working together to develop a machine-learning-based smart alarm system for the CEBAF injector. Because of the injector’s large number of parameters and possible fault scenarios, it is highly desirable to have an autonomous alarm system that can quickly identify and diagnose unusual machine states. We present our work on artificial neural networks designed to identify such undesirable machine states. In particular, we test both auto-encoders and inverse models as possible tools for differentiating between normal and abnormal states. These models are being developed using both supervised and unsupervised learning techniques, and are being trained using CEBAF injector data collected during dedicated machine studies as well as during regular operations. Lastly, we discuss tradeoffs between the two types of models.

DOI •

reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-WEPV023

About •

Received ※ 10 October 2021 Accepted ※ 19 January 2022 Issue date ※ 14 March 2022

Cite •

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

WEPV024

X-Ray Beamline Control with Machine Learning and an Online Model

695

B. Nash, D.T. Abell, D.L. Bruhwiler, E.G. Carlin, J.P. Edelen, M.V. Keilman, P. Moeller, R. Nagler, I.V. Pogorelov, S.D. Webb
RadiaSoft LLC, Boulder, Colorado, USA
Y. Du, A. Giles, J. Lynch, J. Maldonado, M.S. Rakitin, A. Walter
BNL, Upton, New York, USA

Funding: This material is based upon work supported by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences under contract DE-SC0020593.
We present recent developments on control of x-ray beamlines for synchrotron light sources. Effective models of the x-ray transport are updated based on diagnostics data, and take the form of simplified physics models as well as learned models from scanning over mirror and slit configurations. We are developing this approach to beamline control in collaboration with several beamlines at the NSLS-II. By connecting our online models to the Blue-Sky framework, we enable a convenient interface between the operating machine and the model that may be applied to beamlines at multiple facilities involved in this collaborative software development.

DOI •

reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-WEPV024

About •

Received ※ 10 October 2021 Accepted ※ 21 November 2021 Issue date ※ 17 December 2021

Cite •

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

WEPV025

Initial Studies of Cavity Fault Prediction at Jefferson Laboratory

700

L.S. Vidyaratne, A. Carpenter, R. Suleiman, C. Tennant, D.L. Turner
JLab, Newport News, Virginia, USA
K.M. Iftekharuddin, M. Rahman
ODU, Norfolk, Virginia, USA

Funding: This work supported by the U.S. Department of Energy, Office of Science, Office of Nuclear Physics under Contract No. DE-AC05-06OR23177.
The Continuous Electron Beam Accelerator Facility (CEBAF) at Jefferson Laboratory is a CW recirculating linac that utilizes over 400 superconducting radio-frequency (SRF) cavities to accelerate electrons up to 12 GeV through 5-passes. Recent work has shown that, given RF signals from a cavity during a fault as input, machine learning approaches can accurately classify the fault type. In this paper we report on initial results of predicting a fault onset using only data prior to the failure event. A data set was constructed using time-series data immediately before a fault (’unstable’) and 1.5 seconds prior to a fault (’stable’) gathered from over 5,000 saved fault events. The data was used to train a binary classifier. The results gave key insights into the behavior of several fault types and provided motivation to investigate whether data prior to a failure event could also predict the type of fault. We discuss our method using a sliding window approach and report on initial results. Recent modifications to the low-level RF control system will provide access to streaming signals and we outline a path forward for leveraging deep learning on streaming data

Poster WEPV025 [1.111 MB]

DOI •

reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-WEPV025

About •

Received ※ 08 October 2021 Revised ※ 19 October 2021 Accepted ※ 11 February 2022 Issue date ※ 05 March 2022

Cite •

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

THAL01

Machine Learning Tools Improve BESSY II Operation

784

L. Vera Ramiréz, T. Birke, G. Hartmann, R. Müller, M. Ries, A. Schälicke, P. Schnizer
HZB, Berlin, Germany

At the HZB user facility BESSY II Machine Learning (ML) technologies aim at advanced analysis, automation, explainability and performance improvements for accelerator and beamline operation. The development of these tools is intertwined with improvements of the prediction part of the digital twin instances at BESSY II [*] and the integration into the Bluesky Suite [**,***]. On the accelerator side, several use cases have recently been identified, pipelines designed and models tested. Previous studies applied Deep Reinforcement Learning (RL) to booster current and injection efficiency. RL now tackles a more demanding scenario: the mitigation of harmonic orbit perturbations induced by external civil noise sources. This paper presents methodology, design and simulation phases as well as challenges and first results. Further ML use cases under study are, among others, anomaly detection prototypes with anomaly scores for individual features.
[*] P. Schnizer et. al, IPAC21
[**] D. Allan, T. Caswell, S. Campbell and M. Rakitin, Synchrot. Radiat. News 32 19-22, 2019
[***] W. Smith et. al, this conference

Slides THAL01 [9.849 MB]

DOI •

reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-THAL01

About •

Received ※ 08 October 2021 Revised ※ 24 October 2021 Accepted ※ 21 November 2021 Issue date ※ 29 January 2022

Cite •

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

THAL02

Bayesian Techniques for Accelerator Characterization and Control

791

R.J. Roussel, A.L. Edelen, C.E. Mayes
SLAC, Menlo Park, California, USA
J.P. Gonzalez-Aguilera, Y.K. Kim
University of Chicago, Chicago, Illinois, USA

Funding: This work was supported by the U.S. National Science Foundation under Award No. PHY-1549132, the Center for Bright Beams.
Accelerators and other large experimental facilities are complex, noisy systems that are difficult to characterize and control efficiently. Bayesian statistical modeling techniques are well suited to this task, as they minimize the number of experimental measurements needed to create robust models, by incorporating prior, but not necessarily exact, information about the target system. Furthermore, these models inherently take into account noisy and/or uncertain measurements and can react to time-varying systems. Here we will describe several advanced methods for using these models in accelerator characterization and optimization. First, we describe a method for rapid, turn-key exploration of input parameter spaces using little-to-no prior information about the target system. Second, we highlight the use of Multi-Objective Bayesian optimization towards efficiently characterizing the experimental Pareto front of a system. Throughout, we describe how unknown constraints and parameter modification costs are incorporated into these algorithms.

Slides THAL02 [4.453 MB]

DOI •

reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-THAL02

About •

Received ※ 10 October 2021 Revised ※ 10 November 2021 Accepted ※ 21 November 2021 Issue date ※ 26 December 2021

Cite •

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

THAL03

Machine Learning Based Middle-Layer for Autonomous Accelerator Operation and Control

797

S. Pioli, B. Buonomo, D. Di Giovenale, C. Di Giulio, L.G. Foggetta, G. Piermarini
LNF-INFN, Frascati, Italy
F. Cardelli, P. Ciuffetti
INFN/LNF, Frascati, Italy
V. Martinelli
INFN/LNL, Legnaro (PD), Italy

The Singularity project, led by National Laboratories of Frascati of the National Institute for Nuclear Physics (INFN-LNF), aim to develop automated machine-independent middle-layer to control accelerator operation through machine learning (ML) algorithms like Reinforcement Learning (RL) and Cluster integrated with accelerator’s sub-systems. In this work we will present architecture and of the middle-layer made with main purpose to drive user requests through the control framework backend and allow users to enjoy a better User Experience (UX) handling system performances without facing problems due to the interaction with control system. We will report the strategy to develop autonomous operation control with RL algorithms together with the fault detection capability improved by Clustering approach as breakdown and waveguide and RF cavity thermal stability monitor. Results of the first period of operation of this system, currently operating at the electron-positron LINAC of the Dafne complex in Frascati, autonomously controlling accelerator performance in terms of beam transport, beam current optimization and RF cavity phase-jitter compensation will be reported.

Slides THAL03 [0.960 MB]

DOI •

reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-THAL03

About •

Received ※ 19 October 2021 Accepted ※ 22 December 2021 Issue date ※ 16 February 2022

Cite •

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

THAL04

Machine Learning Based Tuning and Diagnostics for the ATR Line at BNL

803

J.P. Edelen, K. Bruhwiler, E.G. Carlin, C.C. Hall
RadiaSoft LLC, Boulder, Colorado, USA
K.A. Brown, V. Schoefer
BNL, Upton, New York, USA

Funding: This material is based upon work supported by the U.S. Department of Energy, Office of Science, Office of Nuclear Physics under Award Number DE-SC0019682.
Over the past several years machine learning has increased in popularity for accelerator applications. We have been exploring the use of machine learning as a diagnostic and tuning tool for transfer line from the AGS to RHIC at Brookhaven National Laboratory. In our work, inverse models are used to either provide feed-forward corrections for beam steering or as a diagnostic to illuminate quadrupole magnets that have excitation errors. In this talk we present results on using machine learning for beam steering optimization for a range of different operating energies. We also demonstrate the use of inverse models for optical error diagnostics. Our results are from studies that use combine simulation and measurement data.

Slides THAL04 [4.845 MB]

DOI •

reference for this paper ※ https://doi.org/10.18429/JACoW-ICALEPCS2021-THAL04

About •

Received ※ 10 October 2021 Revised ※ 22 October 2021 Accepted ※ 06 February 2022 Issue date ※ 01 March 2022

Cite •

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