ICALEPCS2023 - Table of Session: MO3A (Feedback Systems & Optimisation)

Paper	Title	Page
MO3AO01	Optimisation of the Touschek Lifetime in Synchrotron Light Sources Using Badger	108
	S.M. Liuzzo, N. Carmignani, L.R. Carver, L. Hoummi, D. Lacoste, A. Le Meillour, T.P. Perron, S.M. White ESRF, Grenoble, France I.V. Agapov, M. Böse, J. Keil, L. Malinapresenter, E.S.H. Musa, B. Veglia DESY, Hamburg, Germany A.L. Edelen, P. Raimondi, R.J. Roussel, Z. Zhang SLAC, Menlo Park, California, USA T. Hellert LBNL, Berkeley, California, USA
	Funding: This project has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No. 871072 Badger* is a software designed to easily access several optimizers (simplex, RCDS*, bayesian optimization, etc.) to solve a given multidimensional minimization/maximization task. The Badger software is very flexible and easy to adapt to different facilities. In the framework of the EURIZON European project Badger was used for the EBS and PETRAIII storage rings interfacing with the Tango and TINE control system. Among other tests, the optimisations of Touschek lifetime was performed and compared with the results obtained with existing tools during machine dedicated times. Z. Zhang et al., "Badger: The Missing Optimizer in ACR", doi:10.18429/JACoW-IPAC2022-TUPOST058 ** X. Huang, "Robust simplex algorithm for online optimization", 10.1103/PhysRevAccelBeams.21.104601
DOI •	reference for this paper ※ doi:10.18429/JACoW-ICALEPCS2023-MO3AO01
About •	Received ※ 28 September 2023 — Revised ※ 08 October 2023 — Accepted ※ 13 October 2023 — Issued ※ 27 October 2023
Cite •	reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)

MO3AO02	Implementation of Model Predictive Control for Slow Orbit Feedback Control in MAX IV Accelerators Using PyTango Framework	116
	C. Takahashi, J. Breunlin, A. Freitaspresenter, M. Sjöström MAX IV Laboratory, Lund University, Lund, Sweden P. Giselsson, E. Jensen Gassheld, M. Karlsson Lund University, Lund, Sweden
	Achieving low emittance and high brightness in modern light sources requires stable beams, which are commonly achieved through feedback solutions. The MAX IV light source has two feedback systems, Fast Orbit Feedback (FOFB) and Slow Orbit Feedback (SOFB), operating in overlapping frequency regions. Currently in MAX IV, a general feedback device implemented in PyTango is used for slow orbit and trajectory correction, but an MPC controller for the beam orbit has been proposed to improve system robustness. The controller uses iterative optimisation of the system model, current measurements, dynamic states and system constraints to calculate changes in the controlled variables. The new device implements the MPC model according to the beam orbit response matrix, subscribes to change events on all beam position attributes and updates the control signal given to the slow magnets with a 10 Hz rate. This project aims to improve system robustness and reduce actuator saturation. The use of PyTango simplifies the implementation of the MPC controller by allowing access to high-level optimisation and control packages. This project will contribute to the development of a high-quality feedback control system for MAX IV accelerators.
	Slides MO3AO02 [4.234 MB]
DOI •	reference for this paper ※ doi:10.18429/JACoW-ICALEPCS2023-MO3AO02
About •	Received ※ 05 October 2023 — Revised ※ 09 October 2023 — Accepted ※ 14 November 2023 — Issued ※ 19 December 2023
Cite •	reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)

MO3AO03	Commissioning and Optimization of the SIRIUS Fast Orbit Feedback	123
	D.O. Tavares, M.S. Aguiar, F.H. Cardoso, E.P. Coelho, G.R. Cruz, A.F. Giachero, L. Lin, S.R. Marques, A.C.S. Oliveira, G.S. Ramirez, É.N. Rolimpresenter, L.M. Russo, F.H. de Sá LNLS, Campinas, Brazil
	The Sirius Fast Orbit Feedback System (FOFB) entered operation for users in November 2022. The system design aimed at minimizing the overall feedback loop delay, understood as the main performance bottleneck in typical FOFB systems. Driven by this goal, the loop update rate was chosen as high as possible, real-time processing was entirely done in FPGAs, BPMs and corrector power supplies were tightly integrated to the feedback controllers in MicroTCA crates, a small number of BPMs was included in the feedback loop and a dedicated network engine was used. These choices targeted a disturbance rejection crossover frequency of 1 kHz. To deal with the DC currents that build up in the fast orbit corrector power supplies, a method to transfer the DC control effort to the Slow Orbit Feedback System (SOFB) running in parallel was implemented. This contribution gives a brief overview of the system architecture and modelling, and reports on its commissioning, system identification and feedback loop optimization during its first year of operation.
	Slides MO3AO03 [78.397 MB]
DOI •	reference for this paper ※ doi:10.18429/JACoW-ICALEPCS2023-MO3AO03
About •	Received ※ 06 October 2023 — Revised ※ 09 October 2023 — Accepted ※ 14 November 2023 — Issued ※ 03 December 2023
Cite •	reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)

MO3AO04	Modelling and Control of a MeerKAT Antenna	131
	I.A. Dodia SARAO, Cape Town, South Africa
	This paper presents a comprehensive approach to modeling for control system design for a MeerKAT antenna. It focuses on dynamic modeling using time and frequency domain techniques, and lays the foundation for the design of a control system to meet the telescope’s stringent pointing and tracking requirements. The paper scope includes rigid body modelling of the antenna, system identification to obtain model parameters, and building a system model in Simulink. The Simulink model allows us to compare model performance with the measured antenna pointing, under various environmental conditions. The paper also integrates models for pointing disturbances, such as wind and friction. The integrated model is compared to the existing control setup. Wind disturbance plays a significant role in the pointing performance of the antenna, therefore the focus is placed on developing an appropriate wind model. This research will conclude by providing a well-documented, systematic control system design that is owned by SARAO and can be implemented to improve the pointing performance of the telescope.
	Slides MO3AO04 [6.441 MB]
DOI •	reference for this paper ※ doi:10.18429/JACoW-ICALEPCS2023-MO3AO04
About •	Received ※ 06 October 2023 — Revised ※ 07 October 2023 — Accepted ※ 14 November 2023 — Issued ※ 18 November 2023
Cite •	reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)

MO3AO05	Path to Ignition at National Ignition Facility (NIF): The Role of the Automated Alignment System	138
	B.P. Patel, A.A.S. Awwal, M. Fedorov, R.R. Leach Jr., R.R. Lowe-Webb, V.J. Miller Kamm, P.K. Singhpresenter LLNL, Livermore, California, USA
	Funding: This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore National Laboratory under Contract DE-AC52-07NA27344 The historical breakthrough experiment at the National Ignition Facility (NIF) produced fusion ignition in a laboratory for the first time and made headlines around the world. This achievement was the result of decades of research, thousands of people, and hardware and software systems that rivaled the complexity of anything built before. The NIF laser Automatic Alignment (AA) system has played a major role in this accomplishment. Each high yield shot in the NIF laser system requires all 192 laser beams to arrive at the target within 30 picoseconds and be aligned within 50 microns-half the diameter of human hair-all with the correct wavelength and energy. AA makes it possible to align and fire the 192 NIF laser beams efficiently and reliably several times a day. AA is built on multiple layers of complex calculations and algorithms that implement data and image analysis to position optical devices in the beam path in a highly accurate and repeatable manner through the controlled movement of about 66,000 control points. The system was designed to have minimum or no human intervention. This paper will describe AA’s evolution, its role in ignition, and future modernization. LLNL Release Number: LLNL-ABS-847783
	Slides MO3AO05 [10.417 MB]
DOI •	reference for this paper ※ doi:10.18429/JACoW-ICALEPCS2023-MO3AO05
About •	Received ※ 22 September 2023 — Revised ※ 07 October 2023 — Accepted ※ 14 November 2023 — Issued ※ 05 December 2023
Cite •	reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)

MO3AO06	Energy Consumption Optimisation by Using Advanced Control Algorithms	145
	F. Ghawash, E. Blanco Viñuela, B. Schofieldpresenter CERN, Meyrin, Switzerland
	Large industries operate energy-intensive equipment and energy efficiency is an important objective when trying to optimize the final energy consumption. CERN utilizes a large amount of electrical energy to run its accelerators, detectors and test facilities, with a total yearly consumption of 1.3 TWh and peaks of about 200 MW. Final energy consumption reduction can be achieved by dedicated technical solutions and advanced automation technologies, especially those based on optimization algorithms, have revealed a crucial role not only in keeping the processes within required safety and operational conditions but also in incorporating financial factors. MBPC (Model-Based Predictive Control) is a feedback control algorithm which can naturally integrate the capability of achieving reduced energy consumption when including economic factors in the optimization formulation. This paper reports on the experience gathered when applying non-linear MBPC to some of the contributors to the electricity bill at CERN: the cooling and ventilation plants (i.e. cooling towers, chillers, and air handling units). Simulation results with cooling towers showed significant performance improvements and energy savings close to 20% over conventional heuristic solutions. The control problem formulation, the control strategy validation using a digital twin and the initial results in a real industrial plant are reported together with the experience gained implementing the algorithm in industrial controllers.
	Slides MO3AO06 [3.101 MB]
DOI •	reference for this paper ※ doi:10.18429/JACoW-ICALEPCS2023-MO3AO06
About •	Received ※ 04 October 2023 — Revised ※ 09 October 2023 — Accepted ※ 14 November 2023 — Issued ※ 29 November 2023
Cite •	reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)

MO3AO07	Control Design Optimisations of Robots for the Maintenance and Inspection of Particle Accelerators	153
	A. Díaz Rosales, M. Di Castropresenter, H. Gamper CERN, Meyrin, Switzerland
	Automated maintenance and inspection systems have become increasingly important over the last decade for the availability of the accelerators at CERN. This is mainly due to improvements in robotic perception, control and cognition and especially because of the rapid advancement in artificial intelligence. The robotic service at CERN performed the first interventions in 2014 with robotic solutions from external companies. However, it soon became clear that a customized platform needed to be developed in order to satisfy the needs and in order to efficiently navigate through the cluttered, semi-structured environment. This led to the formation of a robotic fleet of about 20 different robotic systems that are currently active at CERN. In order to increase the efficiency and robustness of robotic platforms for future accelerators it is necessary to consider robotic interventions at the early design phase of such machines. Task specific solutions tailored to the specific needs can then be designed, which in general show higher efficiency than multipurpose industrial robotic systems. This paper presents current advances in the design and development of task specific robotic system for maintenance and inspection in particle accelerators, taking the 100 km long Future Circular Collider main tunnel as a use case. The requirements on such a robotic system, including the applied control strategies, are shown, as well as the optimization of the topology and geometry of the robotic system itself.
	Slides MO3AO07 [3.560 MB]
DOI •	reference for this paper ※ doi:10.18429/JACoW-ICALEPCS2023-MO3AO07
About •	Received ※ 29 September 2023 — Revised ※ 10 October 2023 — Accepted ※ 14 November 2023 — Issued ※ 26 November 2023
Cite •	reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)

Paper

Title

Page

Optimisation of the Touschek Lifetime in Synchrotron Light Sources Using Badger

108

S.M. Liuzzo, N. Carmignani, L.R. Carver, L. Hoummi, D. Lacoste, A. Le Meillour, T.P. Perron, S.M. White
ESRF, Grenoble, France
I.V. Agapov, M. Böse, J. Keil, L. Malinapresenter, E.S.H. Musa, B. Veglia
DESY, Hamburg, Germany
A.L. Edelen, P. Raimondi, R.J. Roussel, Z. Zhang
SLAC, Menlo Park, California, USA
T. Hellert
LBNL, Berkeley, California, USA

Funding: This project has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No. 871072
Badger* is a software designed to easily access several optimizers (simplex, RCDS**, bayesian optimization, etc.) to solve a given multidimensional minimization/maximization task. The Badger software is very flexible and easy to adapt to different facilities. In the framework of the EURIZON European project Badger was used for the EBS and PETRAIII storage rings interfacing with the Tango and TINE control system. Among other tests, the optimisations of Touschek lifetime was performed and compared with the results obtained with existing tools during machine dedicated times.
* Z. Zhang et al., "Badger: The Missing Optimizer in ACR", doi:10.18429/JACoW-IPAC2022-TUPOST058
** X. Huang, "Robust simplex algorithm for online optimization", 10.1103/PhysRevAccelBeams.21.104601

DOI •

reference for this paper ※ doi:10.18429/JACoW-ICALEPCS2023-MO3AO01

About •

Received ※ 28 September 2023 — Revised ※ 08 October 2023 — Accepted ※ 13 October 2023 — Issued ※ 27 October 2023

Cite •

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

MO3AO02

Implementation of Model Predictive Control for Slow Orbit Feedback Control in MAX IV Accelerators Using PyTango Framework

116

C. Takahashi, J. Breunlin, A. Freitaspresenter, M. Sjöström
MAX IV Laboratory, Lund University, Lund, Sweden
P. Giselsson, E. Jensen Gassheld, M. Karlsson
Lund University, Lund, Sweden

Achieving low emittance and high brightness in modern light sources requires stable beams, which are commonly achieved through feedback solutions. The MAX IV light source has two feedback systems, Fast Orbit Feedback (FOFB) and Slow Orbit Feedback (SOFB), operating in overlapping frequency regions. Currently in MAX IV, a general feedback device implemented in PyTango is used for slow orbit and trajectory correction, but an MPC controller for the beam orbit has been proposed to improve system robustness. The controller uses iterative optimisation of the system model, current measurements, dynamic states and system constraints to calculate changes in the controlled variables. The new device implements the MPC model according to the beam orbit response matrix, subscribes to change events on all beam position attributes and updates the control signal given to the slow magnets with a 10 Hz rate. This project aims to improve system robustness and reduce actuator saturation. The use of PyTango simplifies the implementation of the MPC controller by allowing access to high-level optimisation and control packages. This project will contribute to the development of a high-quality feedback control system for MAX IV accelerators.

Slides MO3AO02 [4.234 MB]

DOI •

reference for this paper ※ doi:10.18429/JACoW-ICALEPCS2023-MO3AO02

About •

Received ※ 05 October 2023 — Revised ※ 09 October 2023 — Accepted ※ 14 November 2023 — Issued ※ 19 December 2023

Cite •

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

MO3AO03

Commissioning and Optimization of the SIRIUS Fast Orbit Feedback

123

D.O. Tavares, M.S. Aguiar, F.H. Cardoso, E.P. Coelho, G.R. Cruz, A.F. Giachero, L. Lin, S.R. Marques, A.C.S. Oliveira, G.S. Ramirez, É.N. Rolimpresenter, L.M. Russo, F.H. de Sá
LNLS, Campinas, Brazil

The Sirius Fast Orbit Feedback System (FOFB) entered operation for users in November 2022. The system design aimed at minimizing the overall feedback loop delay, understood as the main performance bottleneck in typical FOFB systems. Driven by this goal, the loop update rate was chosen as high as possible, real-time processing was entirely done in FPGAs, BPMs and corrector power supplies were tightly integrated to the feedback controllers in MicroTCA crates, a small number of BPMs was included in the feedback loop and a dedicated network engine was used. These choices targeted a disturbance rejection crossover frequency of 1 kHz. To deal with the DC currents that build up in the fast orbit corrector power supplies, a method to transfer the DC control effort to the Slow Orbit Feedback System (SOFB) running in parallel was implemented. This contribution gives a brief overview of the system architecture and modelling, and reports on its commissioning, system identification and feedback loop optimization during its first year of operation.

Slides MO3AO03 [78.397 MB]

DOI •

reference for this paper ※ doi:10.18429/JACoW-ICALEPCS2023-MO3AO03

About •

Received ※ 06 October 2023 — Revised ※ 09 October 2023 — Accepted ※ 14 November 2023 — Issued ※ 03 December 2023

Cite •

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

MO3AO04

Modelling and Control of a MeerKAT Antenna

131

I.A. Dodia
SARAO, Cape Town, South Africa

This paper presents a comprehensive approach to modeling for control system design for a MeerKAT antenna. It focuses on dynamic modeling using time and frequency domain techniques, and lays the foundation for the design of a control system to meet the telescope’s stringent pointing and tracking requirements. The paper scope includes rigid body modelling of the antenna, system identification to obtain model parameters, and building a system model in Simulink. The Simulink model allows us to compare model performance with the measured antenna pointing, under various environmental conditions. The paper also integrates models for pointing disturbances, such as wind and friction. The integrated model is compared to the existing control setup. Wind disturbance plays a significant role in the pointing performance of the antenna, therefore the focus is placed on developing an appropriate wind model. This research will conclude by providing a well-documented, systematic control system design that is owned by SARAO and can be implemented to improve the pointing performance of the telescope.

Slides MO3AO04 [6.441 MB]

DOI •

reference for this paper ※ doi:10.18429/JACoW-ICALEPCS2023-MO3AO04

About •

Received ※ 06 October 2023 — Revised ※ 07 October 2023 — Accepted ※ 14 November 2023 — Issued ※ 18 November 2023

Cite •

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

MO3AO05

Path to Ignition at National Ignition Facility (NIF): The Role of the Automated Alignment System

138

B.P. Patel, A.A.S. Awwal, M. Fedorov, R.R. Leach Jr., R.R. Lowe-Webb, V.J. Miller Kamm, P.K. Singhpresenter
LLNL, Livermore, California, USA

Funding: This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore National Laboratory under Contract DE-AC52-07NA27344
The historical breakthrough experiment at the National Ignition Facility (NIF) produced fusion ignition in a laboratory for the first time and made headlines around the world. This achievement was the result of decades of research, thousands of people, and hardware and software systems that rivaled the complexity of anything built before. The NIF laser Automatic Alignment (AA) system has played a major role in this accomplishment. Each high yield shot in the NIF laser system requires all 192 laser beams to arrive at the target within 30 picoseconds and be aligned within 50 microns-half the diameter of human hair-all with the correct wavelength and energy. AA makes it possible to align and fire the 192 NIF laser beams efficiently and reliably several times a day. AA is built on multiple layers of complex calculations and algorithms that implement data and image analysis to position optical devices in the beam path in a highly accurate and repeatable manner through the controlled movement of about 66,000 control points. The system was designed to have minimum or no human intervention. This paper will describe AA’s evolution, its role in ignition, and future modernization.
LLNL Release Number: LLNL-ABS-847783

Slides MO3AO05 [10.417 MB]

DOI •

reference for this paper ※ doi:10.18429/JACoW-ICALEPCS2023-MO3AO05

About •

Received ※ 22 September 2023 — Revised ※ 07 October 2023 — Accepted ※ 14 November 2023 — Issued ※ 05 December 2023

Cite •

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

MO3AO06

Energy Consumption Optimisation by Using Advanced Control Algorithms

145

F. Ghawash, E. Blanco Viñuela, B. Schofieldpresenter
CERN, Meyrin, Switzerland

Large industries operate energy-intensive equipment and energy efficiency is an important objective when trying to optimize the final energy consumption. CERN utilizes a large amount of electrical energy to run its accelerators, detectors and test facilities, with a total yearly consumption of 1.3 TWh and peaks of about 200 MW. Final energy consumption reduction can be achieved by dedicated technical solutions and advanced automation technologies, especially those based on optimization algorithms, have revealed a crucial role not only in keeping the processes within required safety and operational conditions but also in incorporating financial factors. MBPC (Model-Based Predictive Control) is a feedback control algorithm which can naturally integrate the capability of achieving reduced energy consumption when including economic factors in the optimization formulation. This paper reports on the experience gathered when applying non-linear MBPC to some of the contributors to the electricity bill at CERN: the cooling and ventilation plants (i.e. cooling towers, chillers, and air handling units). Simulation results with cooling towers showed significant performance improvements and energy savings close to 20% over conventional heuristic solutions. The control problem formulation, the control strategy validation using a digital twin and the initial results in a real industrial plant are reported together with the experience gained implementing the algorithm in industrial controllers.

Slides MO3AO06 [3.101 MB]

DOI •

reference for this paper ※ doi:10.18429/JACoW-ICALEPCS2023-MO3AO06

About •

Received ※ 04 October 2023 — Revised ※ 09 October 2023 — Accepted ※ 14 November 2023 — Issued ※ 29 November 2023

Cite •

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

MO3AO07

Control Design Optimisations of Robots for the Maintenance and Inspection of Particle Accelerators

153

A. Díaz Rosales, M. Di Castropresenter, H. Gamper
CERN, Meyrin, Switzerland

Automated maintenance and inspection systems have become increasingly important over the last decade for the availability of the accelerators at CERN. This is mainly due to improvements in robotic perception, control and cognition and especially because of the rapid advancement in artificial intelligence. The robotic service at CERN performed the first interventions in 2014 with robotic solutions from external companies. However, it soon became clear that a customized platform needed to be developed in order to satisfy the needs and in order to efficiently navigate through the cluttered, semi-structured environment. This led to the formation of a robotic fleet of about 20 different robotic systems that are currently active at CERN. In order to increase the efficiency and robustness of robotic platforms for future accelerators it is necessary to consider robotic interventions at the early design phase of such machines. Task specific solutions tailored to the specific needs can then be designed, which in general show higher efficiency than multipurpose industrial robotic systems. This paper presents current advances in the design and development of task specific robotic system for maintenance and inspection in particle accelerators, taking the 100 km long Future Circular Collider main tunnel as a use case. The requirements on such a robotic system, including the applied control strategies, are shown, as well as the optimization of the topology and geometry of the robotic system itself.

Slides MO3AO07 [3.560 MB]

DOI •

reference for this paper ※ doi:10.18429/JACoW-ICALEPCS2023-MO3AO07

About •

Received ※ 29 September 2023 — Revised ※ 10 October 2023 — Accepted ※ 14 November 2023 — Issued ※ 26 November 2023

Cite •

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