EPICS Collaboration Meeting Fall 2026

1 Nov 2026, 06:00 → 6 Nov 2026, 18:00 Asia/Shanghai

Dongguan Exhibition International Hotel

The Experimental Physics and Industrial Control System, better known as EPICS, is an open-source software framework widely used by scientific facilities around the world to build distributed control systems for large-scale experimental instruments. It supports applications in areas such as particle accelerators, telescopes, experimental beamlines, neutron sources, and other complex scientific platforms. The EPICS Collaboration Meetings bring together developers, engineers, users, and project managers from EPICS sites worldwide to exchange ideas, share practical experience, discuss ongoing developments, and coordinate future plans. These meetings also help the community learn from one another, improve existing tools, explore new solutions, and reduce duplicated effort across different facilities.

The Fall 2026 EPICS Collaboration Meeting will take place in Dongguan, Guangdong, China, from November 1 to 6, 2026. The event is hosted by the Institute of High Energy Physics (IHEP) and organized by the China Spallation Neutron Source (CSNS).Check-in and on-site registration will be available on Sunday, November 1. Technical sessions will run from Monday, November 2, through the morning of Wednesday, November 4. A visit to the CSNS facility is scheduled for Wednesday afternoon, followed by the conference banquet that evening. Training sessions will take place on Thursday and Friday, November 5–6.Please refer to the timetable for the detailed programme and latest updates.

We warmly look forward to welcoming you to Dongguan and sharing a productive and enjoyable week together at CSNS.

Please find links to previous meetings.

 

 

Sunday, 1 November

Registration

Monday, 2 November

Status reports

09:00 Welcome remarks

1 CSNS: Current Status, Upgrades, and Future Developments

Speaker: Yuliang Zhang (Institute of High Energy Physics)

2 History of EPICS

Speaker: Bob Dalesio

3 EPICS Applications@NSRL

Speaker: Gongfa Liu

10:20 Coffe Break

4 EPICS-based Control Systems at SARI

Shanghai Advanced Research Institute (SARI), Chinese Academy of Sciences (CAS) focuses on the development, construction, and operation of large advanced light source facilities, including the Shanghai Synchrotron Radiation Facility (SSRF), Shanghai soft X-ray Free Electron Laser facility (SXFEL), and Shanghai HIgh repetition rate XFEL aNd Extreme light facility (SHINE). Extremely stringent requirements are imposed by these facilities on beam instrumentation, control and timing. The complete technical chain covering EPICS-based control system, high-precision timing distribution, beam instrumentation, high-speed signal acquisition and processing electronics have been established. The core technical achievements, system architecture, and key performance indicators will be summarized in this report.

Speaker: 映炳 阎 (中国科学院上海高等研究院)

5 Controls ESS

Building a project with no hesitancy to change everything over and over.

Speaker: Sky Brewer (ESS)

6 EPICS-Based Control Systems at IHEP: From CSNS to CSNS-II

Speaker: Dr 晓含（Xiaohan） 卢 (Lu) (高能所)

Lunch

Experiment control, data acquisition, data analysis, AI/ML

7 The Beamline Control System of HEPS

The High Energy Photon Source (HEPS) is China's first fourth-generation high-energy synchrotron radiation facility, designed to operate at 6 GeV with a natural emittance below 60 pm·rad. Its beamline control system is responsible for high-precision automated control, status monitoring, and data acquisition across all optical components, diagnostic devices, and sample stages, ensuring the delivery of stable, reliable, and high-performance photon beams to experimental end-stations. This article presents the architecture, hardware, software, and operational status of the HEPS beamline control system, which is built on the EPICS framework and features a three-tier architecture, a three-network topology, and a standardized motion control system.

Speaker: 刚 李 (高能所)

8 Modernizing Accelerator Application Development with AI

Rapid advances in AI are reshaping software development through agentic workflows, significantly improving productivity and enabling previously impractical tasks. This talk will present our efforts to apply these technologies to application development, with approaches and lessons that may also benefit the broader accelerator community.

Speaker: Tong Zhang (Facility for Rare Isotope Beams, Michigan State University)

9 SuperKEKB timing system

The SuperKEKB timing system has a unique configuration. It consists of both Event Timing System and White Rabbit. Besides, we utilize Event modules provided by MRF and SARI. We realized the complicated operation of SuperKEKB by making use of the characteristics of those modules.

Speaker: Hiroshi Kaji (KEK)

10 The Design and Operation of SHINE Machine Protection System

Shanghai HIgh repetitioN rate XFEL and Extreme light facility (SHINE) interlock system comprises approximately 100 equipment cabinets and 10 000 interlock hardware signals. With the ongoing installation and commissioning of the SHINE facility, the scale of signals integrated into the interlock system continues to expand. This study aims to improve the efficiency of system development and maintenance, as well as to enable rapid fault localization and identification, while fulfilling the basic protection functions. This research focused on the SHINE interlock system. The hardware design, software templates, and signal interface protocols were standardized. An automated generation approach was implemented to produce interlock software for both the Input/Output Controller (IOC) layer and operator interface layer. Furthermore, a top-down interface design was adopted, integrating multi-dimensional data, including sector-specific views, subsystem statuses, comprehensive machine parameters, and visual/auditory alarms, into a unified supervisory framework. The development and implementation of these functionalities significantly reduce the workload of repetitive coding, and the automated generation process achieve a 100% correctness rate in EPICS control system development. Fault indication and diagnostic capabilities are substantially improved, enabling localization of typical interlock events within seconds. The final deployed system manages all 10 000 hardware signals with full functional coverage. Currently, the automated interlock program and the overall interface design have been completed and successfully deployed in SHINE.

Speaker: 春蕾 于 (中国科学院上海高等研究院)

11 CSNS Target Station Control System Based on EPCIS

It introduces the overall layout of the target station control system for CSNS Phase I, the EPICS access methods of each subsystem, and the main problems encountered during operation, as well as a brief introduction to the upgrade of CSNS Phase II.

Speaker: Jun Xu (中国科学院高能物理研究所)

12 The Control System of High Current Multi-Pulse Linear Induction Accelerator

The high-current multi-pulse linear induction accelerator operates in the burst mode. During a single operation, the control system must execute output control for multiple power supply groups, coordinate subsystem scheduling, manage timing triggers, implement fast protection, and monitor the entire process with software braking. Designed with a three-layer standard model architecture based on EPICS, the control system develops prototype power supply group controls, an integrated master control platform, and timing trigger and fast protection system for critical burst-mode operations. This paper details the operational workflow, overall architecture, hardware and software platform design, and key subsystem development of high-current LIA control systems, concluding with future research directions.

Speaker: 泽同 李 (中国工程物理研究院流体物理研究所)

13 Design of an EPICS-Integrated Digital Twin Architecture with Multi-Physics Model Services for the NJU BNCT Facility

Accelerator-based Boron Neutron Capture Therapy (AB-BNCT) requires coordinated monitoring and safety evaluation across proton beam delivery, high-heat-flux target cooling, and the beam shaping assembly (BSA). While high-fidelity numerical simulations—such as MCNP6 for radiation transport and COMSOL Multiphysics for conjugate heat transfer—provide detailed physical insights, single-case evaluations typically require tens of minutes to hours, limiting their direct use in online operational assessment.
In this contribution, we present the conceptual design and prototyping progress of a layered EPICS-integrated Digital Twin framework for the BNCT facility under development at Nanjing University. Rather than replacing standard control protocols, the proposed architecture utilizes EPICS Channel Access / PVAccess (CA/PVA) and Python-based virtual IOCs to orchestrate decoupled back-end physics model services.
Specifically, for the target station, we model a 14 MeV proton beam impinging on a stratified Be-Li composite target backed by a spiral water-cooled copper heat sink. Offline multi-physics datasets—capturing 3D Bragg-peak energy deposition under various beam spot profiles (e.g., uniform vs. Gaussian distributions) and spiral-channel thermal-hydraulic performance—are structured to build rapid-evaluation thermal surrogate models, while downstream BSA neutron spectrum estimation interfaces with the group's physically constrained Energy Spectrum Superposition Method–Machine Learning (ESSM-ML) algorithm.
By subscribing to both upstream beam parameters (current, energy, spot profile) and coolant loop PVs (flow rate, inlet temperature, pressure drop), the twin service provides model-based virtual diagnostics for internal states that are difficult to measure directly online, such as peak lithium and wet-wall temperatures against melting and cavitation limits.
In the current pre-commissioning stage, this framework supports non-intrusive virtual commissioning and parasitic (shadow-mode) software testing for target thermal-safety alarm logic and beam-target matching prior to hardware integration.

Speaker: Xia Yan (Nanjing University)

14 Implementation of IOC Access Security Management in Particle Accelerator

In particle accelerator control systems, taking CAFeII and HIAF as examples, the accelerator uses mode management to dynamically rebuild device operation safety boundaries in different operating scenarios. It defines mode objects and configuration information, and IOC access security management is used to enforce access security constraints for mode objects under different modes. IOC access security management can implement permission restrictions from various dimensions, including UAG (User Access Group), HAG (Host Access Group), ASG (Access Security Group), ASL (Access Security Level), and IPAG (IP Access Group), which are implied in CAFeII and HIAF. CAFeII and HIAF use UAG, ASL, ASG, and IPAG. Mode management sets up an operation permission matrix, and IOC permission management uses this to implement a user-role-based control matrix, which specifies different user roles’ read, write, modify, and force-overwrite permissions on specific parameter sets in different operating modes. Different operating modes involve different accelerator systems, and the same system may have different parameter configurations under different modes. Parameter sets are divided by system and correspond to different array PVs. By checking the configuration matrix against the corresponding array index numbers and version numbers, write permissions for consistent configuration parameters are granted. IOC access security management ensures precise control when switching accelerator operation modes, prevents human error, and is the core of accelerator mode management.

Speaker: Ms Jiaosai Li (ADVANCED ENERGY SCIENCE AND TECHNOLOGY GUANGDONG LABORATORY)

15:40 赞助商中能聚控

15:55 Coffee Break

15 Design and Implementation of Hierarchical Power Supply Control System for HEPS

This paper presents the power supply control system developed for the High Energy Photon Source (HEPS). Over 2500 power supplies at 58 stations on the 1346 m storage ring are controlled via a hierarchical EPICS IOC structure with custom-built Modbus drivers. In the Booster, each power supply(Dynamic Ps) IOC runs in an isolated Docker container and the dynamic status data is captured at up to 10 Hz using pySignal and asyncio. The softIOC-based monitoring framework synchronizes data to the central control system and database, and triggers real-time WeCom alerts. Redis is introduced as an in-memory cache to minimize data access latency. The integrated architecture provides an efficient, low-latency and reliable control solution for HEPS power supplies.

Speaker: Dr 艳华 卢 (高能所)

16 Beam diagnostics and its electronics for the CSNS high-current accelerator

The China Spallation Neutron Source (CSNS) is a major facility for neutron science in China, currently operating at an average beam power of 170 kW, with a beam energy of 1.6 GeV and a repetition rate of 25 Hz. In 2024, the CSNS Upgrade Project (CSNS-II) was launched, introducing new instrumentation for the superconducting linac, RCS, and additional beamlines. In 2025, we have upgraded the instrumentations in the RCS injection area featured with the SEM grids with d100μm CNT fibers, and a rectangular linear-cut BPM. This presentation provides an overview of the new diagnostics systems and their associated challenges, such as the non-intercept beam profile monitors at the superconducting linac and RCS, the wide-dynamic range BPM system for the RCS, and the beam-loss monitors for the fast machine protection system. It also reviews recent R&D efforts on advanced technologies, including laser-wire monitors, ionization profile monitors, fast Faraday cup/EO modulation, BPM electronics, PXIe digitizer board, and innovative materials such as carbon nanotube (CNT) wires, fluorescent wires.

Speaker: UNKNOWN 杨仁俊

17 Integration of Multi-Objective Bayesian Optimization with EPICS for Intelligent Beam Tuning at CSNS

We present an EPICS-native Bayesian optimization toolkit for automated beam commissioning at CSNS. Built on the Pyapas framework, the system integrates seamlessly with existing controls via EPICS Channel Access, treating all hardware setpoints and diagnostics as PVs. It supports both single-objective BO and multi-objective BO using qEHVI to optimize beam transmission and loss simultaneously. A layered machine protection scheme—including hard parameter bounds, firmware interlocks, and automatic beam aborts—ensures safe operation. Experimental campaigns across the linac and RCS demonstrate significant improvements in transmission efficiency and beam loss reduction. The modular architecture enables rapid, non-disruptive deployment across accelerator subsystems.

Speaker: Dr Hanyang Liu (IHEP)

18 Development of a Beam Dynamics Simulation Framework for High-Intensity Fixed-Field Alternating-Gradient Accelerators

High-intensity fixed-field alternating-gradient accelerators pose challenges for beam dynamics simulation due to their strongly nonlinear magnetic fields, energy-dependent closed orbits, and space-charge effects. These characteristics complicate the modeling of particle trajectories and collective beam dynamics during injection, acceleration, and extraction.

This presentation introduces a Python-based beam dynamics simulation framework developed to address these challenges. The framework supports particle tracking in nonlinear magnetic fields and incorporates a 2.5D space-charge model based on dynamically reconstructed local coordinate frames, enabling simulations of longitudinally extended beams along curved, energy-dependent reference trajectories. It also provides tools for studying multi-turn injection, transverse phase-space painting, beam accumulation, and extraction trajectories.

Parallel computing is employed for multi-particle simulations and parameter studies. Future developments include GPU acceleration for particle tracking and space-charge calculations, along with automated simulation workflows and artificial intelligence-assisted parameter optimization. Representative simulation results will be presented to illustrate the framework's capabilities.

Speaker: 凯 周 (IHEP)

19 A High-Sampling-Rate Data Acquisition System Based on Timing Triggers

This talk presents the design of a high-sampling-rate data acquisition system based on timing triggers. The system distributes timing trigger signals through a timing master and timing slaves to control data acquisition at the front-end devices. Power supply controllers are used as simulated front-end devices to generate sampled data. The data are buffered in fixed time windows according to the trigger signals and transmitted to a data acquisition computer via the User Datagram Protocol (UDP). The receiving system performs data reception, reordering, and integrity checks before forwarding the data to Apache Kafka for buffering and online distribution. At the storage end, HDF5 is used to organize and store data from multiple devices in a structured format.

Speaker: 丰 邴 (中国科学院高能物理研究所)

20 An EPICS-Based Framework for AI-Assisted Autonomous Beam Tuning of a 2×6 MV Tandem Accelerator

Autonomous beam tuning of a tandem accelerator requires the coordinated control of multiple beamline devices and the continuous processing of beam diagnostic information. In the 2×6 MV tandem accelerator, the beam transport system consists of multiple electrostatic and magnetic elements distributed across several beamlines, resulting in a complex and highly coupled tuning problem. To support the future development of AI-assisted autonomous beam tuning, an integration approach based on the existing EPICS control system is proposed.

The existing EPICS control system provides access to beam diagnostic data, device parameters, and beamline control interfaces. Based on this infrastructure, an AI-assisted tuning framework is proposed as a high-level decision layer that can interact with accelerator devices through EPICS process variables. This approach enables intelligent tuning strategies to be integrated with the existing control infrastructure without requiring fundamental modifications to the underlying machine control system. Historical beam tuning data can provide a basis for developing data-driven models and intelligent tuning strategies. Safety considerations, including parameter validation, operational limits, beam-loss monitoring, and abnormal-operation handling, are incorporated into the proposed architecture to support safe interaction between the AI decision layer and the machine control system.

This work presents an approach for integrating AI-assisted beam tuning with an existing EPICS-based accelerator control system. The proposed framework provides an interface between accelerator physics, machine learning, and accelerator control, and establishes a foundation for future implementation and experimental validation of AI-assisted autonomous beam tuning on the 2×6 MV tandem accelerator.

Speaker: Jian Kang (中国原子能科学研究院)

21 An Intelligent Agent-Assisted System for HEPS Beamline Operations

Beamline operation at the High Energy Photon Source (HEPS) requires the integration of control systems, experimental data, operational knowledge, and data-analysis tools. This work presents an intelligent agent-assisted framework for HEPS beamline operations based on locally deployed Large Language Models (LLMs) and the Osprey agent platform. The framework connects the language model with beamline data and control interfaces while keeping experimental data within the local HEPS computing environment.
A controlled Plan–Check–Execute workflow is adopted to separate LLM reasoning from direct device operations. The current system integrates internal technical documents, EPICS process variables, historical archive data, detector images, and scan-control interfaces. Implemented functions include source-referenced local document retrieval, visualization of historical EPICS data, detector image reconstruction and OpenCV-based image analysis, and one-dimensional motor scanning.
In a representative scanning experiment, detector images acquired at different motor positions are analyzed using image metrics, allowing the system to recommend a refined scan range for subsequent verification and execution. The results demonstrate that the proposed framework can assist users in experimental planning, data analysis, and scan optimization within a locally deployed and controlled environment. Human supervision and confirmation are retained for operations that modify device states, thereby improving operational efficiency while maintaining data security and experimental controllability.

Speaker: Qun Zhang (IHEP)

22 模式驱动的加速器参数快照管理与快速恢复

超导直线加速器涉及大量电源、射频及束流控制设备，运行参数分布于三十多个控制界面中，且不同子系统之间存在状态依赖。为解决超导直线加速器检修、调束及状态切换中参数分散、跨子系统执行顺序复杂和恢复结果难以统一确认等问题，构建了基于实验物理和工业控制系统（EPICS）与Phoebus保存与恢复（SAR）的模式驱动参数快照管理与恢复方法。该方法面向电源、低电平射频、定时、机器保护和束流允许五个子系统，建立目标模式与参数组、状态快照、执行阶段及校验规则的映射，通过分阶段调度和阶段内并行写入完成参数恢复，并结合回读校验、自动重试和异常处理形成闭环。在实际控制环境中开展恢复性能、20次重复执行及异常测试。结果表明，各子系统参数均能按既定阶段完成恢复与校验，多次执行的结果一致且耗时波动较小。该方法实现了多子系统参数恢复的统一组织与确认，提高了模式切换的可控性和异常可追溯性。

Speaker: 苗苗 廖 (中国科学院近代物理研究所)

miaomiao liao.pdf

17:55 赞助商易事特

Tuesday, 3 November

User Interfaces, Client Tools, Operational Tools

23 Scientific Control Systems

Speaker: Bob Dalesio

24 NTTable: A case study of Normative Types

The Normative Types are a standard high-level data types used to represent more complex data than the EPICS V3 Data Types ("DBRs"). On this presentation some the possibilities of using Normative Types, in special NTTables, will be explained. From controlling the timing system to showing versions of used EPICS Modules the experience with NTTables from the European Spallation Source (ESS) is going to be presented. Different approaches, examples and pitfalls will be part of this talk.

Speaker: Gabriel Fedel (European Spallation Source)

25 CSNS Contributions to the EPICS Community

After CSNS went into operation, the control group developed a set of web-based applications for operation and maintenance. This effort accumulated significant software development skills and experience. Building on this foundation, CSNS began contributing to the EPICS community by providing both new tools deployable by other laboratories and source code contributions to existing projects.

These contributions cover database applications, the Node.js ecosystem, web tools, and collaboration on classic repositories. In the database area, we developed clog (a compact electronic logbook), elog-mobile (a mobile web client for PSI Elog), and e-traveler with a modern technology stack. For the Node.js ecosystem, packages like node-epics-ca, node-epics-pcas, and node-epics-pvxs enable web services to interface with EPICS. Our web tools include capva (unifying CA and PVA in Python), epics-rest-ws (exposing REST/WebSocket access), web-pvtools (browser-based PV tools), and archiver-status. Beyond these standalone packages, CSNS has also contributed source code to major projects including Phoebus, the EPICS Archiver Appliance, and pyepics.

This talk will present these packages and patches. All independent tools are available on GitHub under the MIT license or similar open-source licenses.

Speaker: Lin Wang (CSNS / IHEP)

26 go-epics: Bridging Go with Native EPICS CA/PVA Libraries

This short talk introduces go-epics, a Go package that bridges the native EPICS Channel Access (CA) and PVAccess (PVA) C/C++ libraries through cgo. By providing high-level, idiomatic Go APIs while preserving the capabilities of the native libraries, go-epics simplifies the development and deployment of portable, cross-platform applications. It also establishes a foundation for future integration with agentic AI workflows.

Speaker: Tong Zhang (Facility for Rare Isotope Beams, Michigan State University)

27 Using an AI Agent for EPICS IOC Development at VIGAS

Developing an IOC for a new device typically takes two days or more by hand, and most of that time goes into reading and analysing the device protocol. Large language models are good at text analysis and code generation, so LLM and AI agent can help with IOC development. We distilled our IOC development experience at VIGAS into an agent skill that constrains and guides the AI agent's behaviour. Development time drops to half a day. The resulting code follows VIGAS conventions and comes with complete and consistent documentation. The workflow as a whole carries less risk than doing it by hand. We have used the skill for IOC development at VIGAS and in several other projects.

Speaker: 闯 胡 (清华大学)

28 Development of Vacuum Control System for High-Current Linear Induction Accelerator

Abstract: The vacuum system provides the necessary vacuum conditions for the generation, transport, and target impact of particle beams in high-current linear induction accelerators (LIAs). To achieve monitoring and control of nearly a hundred molecular pumps, several cryopumps, and numerous vacuum valves in two high-current LIAs, this paper designs a vacuum control system based on a front-end field controller and a cloud-based EPICS architecture. The front-end field controller enables one-click start/stop of the vacuum systems for the two high-current LIAs and vacuum protection under abnormal conditions through user-defined logic, and connects to the accelerator control network via the TCP/IP protocol. Multiple application-layer services based on the EPICS architecture are developed on virtual hosts of the accelerator cloud platform. Field data is transmitted in real time using the S7Nodave protocol, enabling remote monitoring of the vacuum system status of the two high-current LIAs, as well as application services such as cathode heating and abnormal protection. After testing and online operation, the vacuum control system demonstrates complete functionality and stable operation, ensuring the stable output of the accelerator.
Key words：High-current linear induction accelerator, vacuum system, EPICS, control system

Speaker: Mr Gang Long (流体物理研究所)

29 Porting and Application of Pyapas for the PWFA Beam Commissioning

Plasma acceleration is a novel acceleration principle featuring a high acceleration gradient. The Plasma Wakefield Acceleration (PWFA) facility at the Institute of High Energy Physics (IHEP) comprises two beamlines, Beamline 1 (BL1) and Beamline 2 (BL2). BL1 delivers positron and electron beams produced by the BEPCII linac at energies up to 2 GeV for plasma acceleration experiments. BL2 is a newly constructed linac with an energy of 150 MeV and bunch charge exceeding 5 nC. Both BL1 and BL2 are under beam commissioning.
Pyapas is a Python-based new modular framework for high-level application originally development at HEPS. Its hardware communication relies on the EPICS Channel Access protocol. In this work, we present the porting and adaptation of Pyapas to the PWFA experimental platform, as well as its application in beam commissioning.

Speaker: Dr YaLiang Zhao (高能所)

10:20 Coffee Break

30 Recent Progress in the EPICS-Based Control System for the J-PARC Muon Linac

The J-PARC Muon Linac is an ongoing project at the Materials and Life Science Experimental Facility (MLF). The accelerator consists of several sections, including the USM, RFQ, IH-DTL, DAW, and DLS, which progressively accelerate muons from 5.6 keV to 212 MeV for the muon g-2/EDM experiment. Construction and commissioning of the linac are being carried out step by step, starting from the upstream sections. In 2026, commissioning has progressed to the RFQ section at 0.34 MeV.
The control system therefore needs to operate partially completed accelerator sections while allowing new sections to be added with minimal modification. The system is based on EPICS and uses PLC I/O modules and MOXA Ethernet-to-serial interfaces for device monitoring and control. PLCs are also used for equipment interlocks. The system was used during muon beam commissioning of the USM and RFQ sections in June 2026 and is now being prepared for the next beam commissioning scheduled for November 2026.
This presentation will give an overview of the control-system architecture and implementation, summarize recent commissioning experience, and discuss ongoing work and plans for future extension.

Speaker: MIN YANG (KEK)

31 The high level application for SHINE

This presentation presents a full suite of self‑developed accelerator upper‑level application software for the SHINE facility. To address the limitations of conventional tools including discontinued maintenance, unsatisfactory runtime efficiency and poor cross‑platform adaptability of EDM, CSS and Phoebus, customized OPI development is carried out based on PyDM. Special widgets featuring security verification and password access control are supplemented to form a SHINE‑exclusive PyDM release, whose source code has been contributed to the EPICS open‑source community.

To tackle the defects of the native Archiver web‑based tool, such as the restriction on axis quantity and loss of extreme‑value samples, an in‑house archiving query and analysis software is implemented with the PyQt framework. It enables fuzzy PV search, multi‑Y‑axis and multi‑curve comparison, and multiple plotting strategies, realizing seamless integration between real‑time operator interfaces and historical data query.

To resolve the information overload issue of existing alarm tools, a comprehensive alarm software is developed by integrating signals from Phoebus/Alarms, MPS and device IOCs. It filters faults according to machine operation modes, performs physical‑region localization, refines alarm messages and supports one‑click interface navigation for rapid fault diagnosis. Meanwhile, the SHINE waveform storage system is constructed with unified HDF5 storage specifications and optimized read‑write performance. Precise alignment of multi‑BPM data is achieved via bunchID, and downloads of both raw and aligned datasets are supported. Covering status monitoring, historical data analysis, fault alarming and waveform streaming‑data storage, this software suite provides comprehensive software support for SHINE facility operation and physics commissioning.

Speaker: 慧慧 吕 (中国科学院上海高等研究院)

32 A Unified EPICS/Phoebus Automation Framework for SRF Cavity Commissioning Across HIAF, CIADS, and IP-SAFE

Commissioning and operation of large-scale superconducting RF (SRF) cavity systems involve repetitive and time-consuming procedures, creating a strong demand for automated operational tools. This work presents a suite of applications based on EPICS and Phoebus for automated cavity conditioning, cavity loading, and RF measurement. The applications integrate EPICS control with Phoebus operator interfaces to execute predefined operational procedures, monitor device status in real time, coordinate safety interlocks, and automatically respond to abnormal conditions during cavity commissioning.

The automation framework is currently being deployed across three accelerator projects at the Institute of Modern Physics: the High Intensity Heavy-ion Accelerator Facility (HIAF), the China Initiative Accelerator Driven System (CIADS), and the IP-SAFE facility. By replacing repetitive manual operations with standardized automated workflows, the framework improves operational efficiency, enhances commissioning consistency, and reduces operator workload. Its modular and reusable design also provides a foundation for extending automation to additional commissioning and operational tasks in future SRF facilities.

Speaker: Dr 丽娟 杨 (近代物理研究所)

33 An Online Fault Detection and Automated Compensation Application on the Pyapas Python Control Framework

High availability of superconducting linacs demands that the control system not only
detect cavity faults but also rapidly reconfigure the remaining RF cavities so that beam
operation can continue with minimal performance loss. This paper presents PEE, a
control-system-level application deployed on Pyapas, a Python-based control system
framework, that closes the loop from fault detection to setpoint deployment for a
44-cavity superconducting linac (20 spoke + 24 elliptical cavities).

The architecture separates monitoring, decision, optimization, and actuation into
loosely coupled Python modules coordinated by a Tkinter-based operator console.
A configurable PV monitoring engine (driven by a JSON configuration) periodically
polls machine process variables — beam energy, beam current, BPM averages, and
per-cavity field readbacks — through the Pyapas channel interface, and evaluates
relative/absolute thresholds to raise graded alarms (critical/warning). Alarm
aggregation rules and a trigger engine decide when a compensation cycle is
launched, with a cooldown timer to prevent oscillation. Faulted cavities are
identified by scanning each cavity's field readback PV against its nominal
setpoint; the surviving field fraction is passed directly to the compensation
engine as a measured quantity rather than an operator guess.

Optimization is delegated to an external beam-dynamics code (TraceWin) orchestrated
as a batch client: a genetic algorithm evolves candidate amplitude/phase setpoints,
each individual being evaluated by running the external simulator and scoring the
resulting envelope against the nominal reference (RMSE), with result caching to
avoid duplicate evaluations. The winning setpoint table is staged in the control
system and exposed to the operator for review before a one-click action writes all
44 cavity amplitude and phase setpoint PVs through the Pyapas channel interface,
keeping a human in the loop for machine-protection reasons.

The system degrades gracefully: when the Pyapas runtime or the machine connection
is unavailable, it automatically falls back to a simulation mode, allowing the
full monitoring → decision → optimization → (simulated) deployment chain to be
tested offline — a feature used for commissioning dry-runs and algorithm
regression testing.

We report the design of the PV configuration layer (which decouples the application
from concrete PV naming conventions), the threading and timeout handling between
the monitoring loop and long-running optimization jobs, the operator-confirmation
workflow, and operational experience with fault scenarios including single-cavity
quench and field-emission degradation.

Speaker: 昕苑 冯 (高能所)

34 Operator Tools for the CSNS Rapid Cycling Synchrotron Built on the Pyapas Framework

The China Spallation Neutron Source Rapid Cycling Synchrotron (CSNS-RCS) operates with a demanding acceleration cycle and tight beam-loss constraints, which puts high requirements on its operation tools. We present a set of operator applications developed on Pyapas, the Python-based accelerator physics application framework used at CSNS, HEPS and BEPCII. These applications talk to the EPICS control system through Channel Access and reuse the framework's common building blocks for GUI, data acquisition and online accelerator modelling with Ocelot.

Five applications are highlighted:

RCS Online Tuner (RCDS). A Python port of the Robust Conjugate Direction Search algorithm for online, noise-tolerant optimization of orbit corrector settings, taking the total beam loss (BLM) or the transmission efficiency as the objective. A virtual-accelerator mock mode allows safe development and training without an IOC.

RCS Local Bump. Computes 41-point field-increment waveforms for up to 37 correctors, following the measured Bρ(t) ramp from injection to extraction. The waveforms are superposed onto the existing EPICS AC waveform channels and verified by readback after each write, supporting the creation of stable local bumps for injection and extraction studies.

MIA/ICA. A turn-by-turn BPM analysis tool that parses TBT data files, applies Second-Order Blind Identification (ICA) to separate coupled betatron modes, and overlays the measured tunes on the resonance diagram. Online connection to the Libra turn-by-turn PVs is being added.

RCS BLM Display. A beam-loss-monitor display showing BLM location photos and waveform statistics (Max/Mean/Std) over the R1–R4 sectors, providing quick identification of loss locations during commissioning and operation.

RCS Parameters Display. A machine-parameter monitor covering 32 BPMs, 82 BLMs and the transmission efficiency, with Libra turn-by-turn BPM waveforms. A 500-shot ring buffer provides real-time display, front/back-window comparison and storage in CSV/HDF5 formats, with automatic alerts when the transmission efficiency drops.

These tools demonstrate how a lightweight Python/EPICS framework can turn machine data into physics-based, interactive operation aids, and how advanced optimization and analysis techniques can be introduced into daily accelerator operation with low development cost.

Speaker: Yuwen An (高能所)

35 An Automatic Loading, Monitoring, and Rapid Recovery Method for Large-Scale Superconducting Cavity Arrays

Large-scale superconducting linear accelerators such as HIAF operate nearly a hundred SRF cavities, whose operation is constrained by slow array startup, Lorentz-force-induced dynamic detuning under pulsed beam, and manual fault recovery. This work presents an integrated automatic loading, online monitoring, and rapid recovery method for the low-level radio frequency (LLRF) system. A distributed multi-process architecture is adopted, with an independent process and a seven-state finite-state machine per cavity, standardizing the sequence from open-loop preloading, amplitude-loop locking, and automatic frequency tuning to phase-loop closure. In pulsed mode, the dynamic detuning is identified online from the pulse waveforms and used to drive pulse-to-pulse mechanical tuning. Faults such as cavity-voltage collapse, quench, and severe detuning are detected by multi-parameter criteria; event-driven hierarchical interlock reset, combined with backfilling of historical steady-state parameters, rebuilds the full closed loop without manual intervention. The system has been deployed on all 96 cavities (17 cryomodules) of HIAF: a single global command loads the entire array in about 5 min, compared with about 120 min manually. In a 10 h parallel run on 81 cavities (about 5.83×10⁵ samples), the median amplitude and phase stabilities reached 0.023% and 0.033°, with 100% closed-loop retention, a roughly 95% automatic recovery success rate, and a mean recovery time of about 45 s versus about 5 min for manual recovery.

Speaker: Qizheng Hou (‌Institute of Modern Physics‌)

36 Bridging LLM Agents and Accelerator Operations: Integrating the Osprey Framework with EPICS at CLAPA-II

As laser-driven plasma accelerator facilities grow in complexity, operators face increasing challenges in cross-subsystem control, historical data correlation and rapid troubleshooting. At Peking University’s Compact Laser Plasma Accelerator II (CLAPA-II), we have deployed an intelligent control assistant built upon Osprey—an agentic interface and safety harness framework designed for safety-critical control systems. Our implementation integrates the AI agent into the CLAPA-II control environment across three core dimensions. First, it establishes direct integration with the EPICS control stack, translating natural language operator intents into auditable channel access operations under strict human-in-the-loop safety constraints. Second, the agent connects to the EPICS Archiver Appliance to autonomously retrieve, visualize, and analyze historical PV time-series data for anomaly detection and post-shot diagnostics. Third, we built a specialized knowledge base for the CLAPA-II trigger system to provide the agent with comprehensive foundational knowledge of our timing infrastructure. Applying Osprey to the trigger system effectively bridges the knowledge gap in specialized experimental scenarios, enabling intelligent control and monitoring while assisting operators with routine maintenance, commissioning, and system upgrades. This talk presents the architecture, operational workflows, and initial practical experiences of deploying Osprey at CLAPA-II.

Speaker: Mr 子恒 卢 (北京大学)

Lunch

Hardware, Driver/Device support

37 EP,CICS - Experimental Physics, Chemistry and Industrial Control System - EPICS for the small chemistry lab

EPICS stands for "Experimental Physics and Industrial Control System" but the physics in that name is doing more assuming than it should. Small chemistry laboratories gas-handling rigs, catalytic reactors, surface-science setups face exactly the same operational reality as small physics beamlines: one or two people covering controls part-time, hardware expected to run for 10 years, no dedicated software team, and a strong preference for reusing proven components over building bespoke software.

This talk asks what changes, and what doesn't, when EPICS moves from a beamline to a chemistry bench. Much carries over directly: serial instrument integration via asyn/StreamDevice, a compiled and long-lived IOC as the core control process, Python kept above that layer rather than replacing it.

What's different is the weight chemistry puts on the IOC as a safety layer, not just a data-acquisition layer. Interlocking gas supplies against pressure and leak detection, guarding ignition sources, and closing control loops on reactions rather than just logging them.

Rather than a finished case study, this is a discussion-opener: a survey of where chemistry-lab control genuinely needs new device support or new patterns, and where it's simply relabeling a problem EPICS has already solved.

The talk closes with concrete questions for the collaboration and an invitation to compare notes with anyone already running EPICS on a chemistry experiment.

Speaker: Junkes Heinz (Fritz Haber Institute)

epcics_slides.pdf

38 Development Status of the Event Timing System

The event timing system has been operating reliably in multiple large accelerator facilities worldwide. Recent developments include a network-wide global interlock function driven by special downstream event codes, a priority-queueing mechanism for full-duplex event codes, and a GNSS-synchronized hardware time-stamping function compatible with the MRF timing system. The system has also been ported to the MicroTCA.4 platform, where each timing module reserves two PCIe buses for communication with the master and slave MCHs, supporting future AI applications. Together with magnet power supplies, low-level RF (LLRF) control, beam diagnostics, and interlock systems, among others, the event timing system enables fine-grained control, real-time fault diagnosis, and intelligent, autonomous commissioning and operation of large accelerator facilities.

Speaker: 鸣 刘 (中国科学院上海高等研究院)

39 EPICS Integration of an MRF VME EVR on a Linux CPU without CR/CSR Support

Standard EPICS integration of MRF VME-EVR timing modules relies on VME CR/CSR space for board discovery, base address assignment, and interrupt setup. To address the issue where certain VME CPUs lack hardware support for CR/CSR access in their PCIe-to-VME bridge, an integration approach was implemented to configure the EVR base address and interrupt over the EVR's onboard UDP interface. A low-level epics interface was also implemented for the CPU's Linux VME driver to handle VME address mapping and data access.

Speaker: Mr Di Wang (KEK)

40 HEPS Timing System

The High Energy Photon Source (HEPS) is a 6-GeV fourth-generation synchrotron light source developed by the Institute of High Energy Physics (IHEP), Chinese Academy of Sciences. It employs a novel on-axis swap-out injection scheme. The timing system is a key subsystem that supplies triggers and clocks to both the accelerators and the beamlines and coordinates the entire facility operation, including the injection scheme. The timing software is based on EPICS. This talk will describe the timing logic and architecture, together with the EPICS-based software.

Speaker: Dr Fang Liu (高能所)

41 An ultra-low latency ATCA-based data acquisition and processing platform for fast protection at CSNS-II

The China Spallation Neutron Source Phase II (CSNS-II) imposes stringent requirements on the real-time monitoring and safety of high-intensity proton beams, necessitating microsecond-level response times for the Fast Protection System. This paper presents a high-throughput, ultra-low-latency data acquisition and processing platform designed on the Advanced Telecommunications Computing Architecture (ATCA). To eliminate standard bus protocol overheads, the system utilizes a custom backplane enabling deterministic, point-to-point interaction between interface and core logic boards via high-speed LVDS signaling (250 MHz) and multi-gigabit serial links (up to 10 Gbps).The platform features a heterogeneous dual-FPGA architecture to maximize processing efficiency. A Radio Frequency System-on-Chip (RFSoC) acts as the front-end processor, employing 14-bit, 5 GSPS direct RF sampling to capture the current signals with minimal analog path delay. Concurrently, a Kintex UltraScale (KU060) FPGA serves as the central logic engine, aggregating device status and executing global protection logic through 10 Gbps transceivers. This functional partitioning ensures parallel execution of signal acquisition and complex logic processing without resource contention. Preliminary results demonstrate that the architecture significantly reduces system loop latency, satisfying the strict timing constraints of the CSNS-II Machine Protection System while providing a scalable solution for high-power accelerator diagnostics.

Speaker: UNKNOWN 朱鹏 (高能所)

42 IOC with Base 7 + StreamDevice for Arb Waveform Generator

A complete EPICS IOC based on EPICS Base 7 and StreamDevice has been developed for the 33600A dual-channel function/arbitrary waveform generator with a LAN interface. It is designed for practical applications such as CT calibration and excitation signal generation in beam experiments.

Speaker: ying 赵颖 (高能所)

43 Retrofitting CSNS Utility Controls: Transition from Commercial SCADA to EPICS

This lightning talk introduces the migration project of the utility facility control system at the China Spallation Neutron Source (CSNS). CSNS auxiliary infrastructure including water cooling, compressed air, HVAC and power distribution originally adopted commercial SCADA software for control. Although commercial SCADA supports fast HMI development and mature industrial functions, it brings vendor lock-in, expensive licenses and closed interfaces. The growing number of subsystems makes unified management challenging.
To achieve a unified utility control system, CSNS initiated a phased migration from commercial SCADA to an EPICS distributed control system. We will present the existing commercial SCADA application status and phased replacement strategy, including PLC-EPICS IOC integration, data archiving, and Phoebus-based HMI reconstruction.
The migration proceeds with no interruption to equipment operation. Dual-system coexistence is maintained while O&M staff make the skill transition from commercial SCADA tools to the EPICS workflow.

Speaker: 绍兴 梁绍兴 (散裂中子源科学中心)

shaoxing liang EPICS Collaboration Meeting Fall 2026-梁绍兴.pptx

44 The development of μTCA.4 based BPM electronics system for the CSNS-II RCS

The CSNS-II upgrade will raise the RCS beam power from about 180 kW to 800 kW, imposing stringent requirements on the BPM electronics, in particular a wide input dynamic range and high position accuracy. The system comprises four local stations, each with an MTCA.4 crate hosting eight BPM channels and one timing board; each channel consists of an RTM board and an AMC board. To cover the wide pickup-signal range of about 54 dB (0.16 Vpp at injection to 80 Vpp at extraction), the RTM front-end employs a four-stage attenuation-switching scheme providing up to 34 dB of attenuation, conditioning the 0–80 Vpp input into the ADC dynamic range. The AMC board, based on a Xilinx Zynq SoC with a 16-bit, 125 MSPS ADC, implements the STD algorithm and provides raw-data, turn-by-turn (TBT) and closed-orbit (COD) modes. Control and data-acquisition software was developed based on EPICS. Laboratory tests show an SNR above 73 dB, an isolation above 85 dB, and an ADC dynamic range of ±1 Vrms. The bunch-by-bunch position resolution is better than 50 µm (0.05% R) in both the x and y planes, and the closed-orbit resolution is better than 5 µm, for signal amplitudes above 0.6 Vpp. In October 2026, eight complete BPM systems were batch-produced and successfully deployed for operation.

Speaker: 春杰 谢

45 The camera control system of China Academy Engineering Physics THz free electron laser facility

A centralized control and image analysis system has been developed for 46 industrial GigE cameras used for beam observation at China Academy Engineering Physics THz free electron laser facility (CTFEL). Each camera is operated by an independent EPICS IOC based on areaDetector and ADAravis. Managing this number of independent IOCs introduces operational challenges beyond basic camera integration, particularly reliable start/stop control, accurate status determination, fault isolation, and unified access to image analysis results.
The system employs a layered architecture consisting of EPICS camera IOCs and a supervisory watchdog on Linux, a Phoebus operator interface, and a Python-based image analysis service. Native camera control and image data are accessed through Channel Access, while supervisory and application-level process variables are published through PVAccess using p4p. A six-state lifecycle model—STOPPED, STARTING, RUNNING, STOPPING, FAULT, and OFFLINE—is used for each camera. Before an IOC is started, the corresponding camera is checked for network availability. The RUNNING state is confirmed using IOC PV connectivity and image acquisition activity rather than the return status of a startup script, preventing false-positive operating states when a camera is disconnected or an IOC fails to recover.
From a unified Phoebus interface, operators can select, start, stop, and monitor individual cameras and inspect live images. The analysis service provides dynamic regions of interest, horizontal and vertical projections, Gaussian fitting, centroid estimation, and weighted beam-size calculations. Configurable accumulated-intensity ratios can be applied independently in both axes to reduce the influence of background and long-tail noise.
This contribution presents the system architecture, supervisory state model, CA/PVA integration, deployment experience, and practical considerations for reliable management of large EPICS camera installations.

Speaker: Shuai Ma (中国工程物理研究院应用物理研究所)

46 Development of a Photoelectric Conversion Communication Device for Megavolt-Class Tandem Accelerators

Equipment at the high-voltage terminal of the main acceleration section of a megavolt-grade tandem accelerator operates inside an SF₆‑filled pressure vessel at 0.607‑0.709 MPa (approximately 7 atm). High-voltage fluctuations and spark‑over events inside the accelerator generate intense electromagnetic radiation, and off-the-shelf signal-conversion devices cannot adapt to such harsh conditions. At present, the accelerator adopts a conventional control scheme in which an insulating rod is driven by a ground-potential motor. This scheme suffers from outdated technical approaches, high manufacturing costs for mechanical structures, and an inability to monitor the operating status of terminal-mounted equipment, significantly limiting safe and stable accelerator operation. Taking a 6 MV tandem accelerator as the research object, this paper develops an opto-electronic communication unit adapted to the extreme operating environment of the accelerator via electrical-to-optical signal conversion. Bidirectional communication between field devices and an EPICS‑based host computer is realized through an optical-fiber link. By simulating the accelerator electric‑field environment and developing dedicated signal-processing and protection circuits, precise monitoring and reliable control of various device signals are achieved. Test results show that the opto-electronic communication unit exhibits good compatibility with the 6 MV tandem accelerator, meets the requirements for long-term stable accelerator operation, and enhances the safety and stability of equipment operation.

Speaker: 祥鹏 孟 (中国原子能科学研究院)

15:30 赞助商汇川

15:45 Coffee Break

47 Use of EPICS in the Mu2e Extinction Monitor System

The Mu2e experiment, under construction at Fermilab, will search for charged lepton flavor violation in the coherent conversion of a muon to an electron in the electric field of an aluminum nucleus. The planned sensitivity of the experiment will be achieved using a pulsed proton beam with a time structure that allows rejection of beam-induced backgrounds. The Mu2e Extinction Monitor measures the time structure of the proton beam using a pixel tracking detector to identify particles from the production target that are either in-time and out-of-time with the expected beam pulses. Control and monitoring of this pixel system, which utilizes the FE-I4b readout chip, is performed using custom electronics and uses EPICS as the primary network control interface. This implementation runs under Debian Linux on a single board computer that interfaces with over 30 devices using GPIO signals and multiple I2C busses to provide over 200 process variables. We will describe the hardware, the EPICS drivers developed for the various sensors and interfaces used in the system, and the control and monitoring of these process variables using EPICS.

Speaker: Matthew Jones (Purdue University)

48 Integrating EtherCAT Motion Controllers into EPICS via Modbus TCP

Two EtherCAT‑based motion control systems have been integrated into EPICS via Modbus TCP at the China Spallation Neutron Source (CSNS). The Very Small Angle Neutron Scattering (VSANS) instrument employs Beckhoff hardware paired with BiSS‑C absolute encoders to implement full‑closed‑loop control of more than 60 axes. Laser interferometer measurements verify its linear positioning accuracy of ±2 μm. The Associated Proton Experimental Platform (APEP) beamline adopts fully domestic Inovance PLCs, where the EPICS IOC runs on-board the AC800 controller. This system controls beamline components including collimators, degraders and sample stages, and has operated stably for over five years. This presentation describes the system architecture, hardware configuration, and EPICS integration approach for both implementations.

Speaker: Yongcheng He (高能所)

49 Compact PLC-type Timing Receiver for J-PARC Main Ring

The timing system of J-PARC accelerator complex has been in operation over 20 years. For J-PARC Main Ring, 45 VME-type timing receivers have been utilized since 2008. However, maintenance has now become nearly impossible.
A new PLC-type receiver was designed in 2024 to replace the VME-type ones. First mass production of 20 pieces took place in 2025. In April 2026, the first PLC-type receiver was put into operation to provide timing signals for the injection kicker system. Three additional receivers are installed during the summer maintenance of 2026, to be in operation after November, 2026.
This report describes details of the PLC-type timing receiver, and current deployment status.

Speaker: Dr Norihiko Kamikubota (KEK)

50 Development of CSNS-II LINAC SRF LLRF

To meet the 500 kW target power of CSNS-II upgrade, a digital LLRF system is developed for the superconducting linear accelerator with 324 MHz spoke and 648 MHz elliptical cavities. An RFSoC-based hardware platform is adopted for direct RF sampling, robust control algorithm is implemented to suppress thermal drift, beam loading and Lorentz detuning. Cavity tests verify amplitude stability better than ±0.3% and phase stability within ±0.3°, satisfying the operational specification of the SRF linac.

Speaker: UNKNOWN 谢哲新 (高能所)

51 Vision-Guided Control for Radioactive Isotope Transfer: A Lab-Scale Validation Framework

Handling radioactive isotopes in the CSNS Phase-II project requires high-precision automation to minimize human radiation exposure. The final deployment will feature a heavy-duty robotic arm mounted on a ground track operating in highly constrained, shielded environments. However, as the final hardware is not yet deployed, this abstract presents a lab-scale validation framework using an AUBO 6-DOF collaborative robotic arm and a RealSense depth camera to de-risk software and algorithm development.
Our core work focuses on vision-based pose estimation and Eye-in-Hand visual servoing. To extract stable 3D coordinates from depth images, we implemented multi-frame filtering and fixed working-plane constraints. Furthermore, to address the limited field of view in narrow storage spaces, we developed an iterative, PID-like visual servoing control logic. The robotic arm actively adjusts its pose to “pull” the target back to the image center, ensuring robust target tracking and alignment during the approach phase.
By decoupling perception and execution, the validated algorithms—including object detection, coordinate transformation, and visual servo logic—are hardware-agnostic and directly transferable to the final track-mounted system. This lab-scale methodology significantly reduces deployment risks and provides a scalable foundation for future integration into distributed control architectures.

Speaker: 清阳 张 (中国散裂中子源)

52 Multimodal Tailgating Detection and Personnel Safety Interlock via OPC UA and EPICS at the CLAPA II Laser Accelerator

Personnel Safety Systems (PPS) at laser-plasma accelerator facilities rely on card-access door controllers to protect radiation-controlled areas. A well-known weakness is tailgating: an unauthorized person follows an authorized entrant through the door before it closes, leaving no credential record that the PPS can act on.

This work presents a multimodal tailgating detection system developed for the CLAPA II laser accelerator at Peking University. Door-access events are obtained through an OPC UA server on the access-control network, while an overhead camera feeds a lightweight people-counting detector running on an edge device (NVIDIA Jetson Nano) on the control network. A fusion state machine aligns the two event streams within a door-open window and compares the authorized count from access logs with the visually observed pass count; a mismatch beyond hysteresis thresholds is raised as a tailgating alarm, corroborated over consecutive windows to suppress false positives. Conditional image recording stores frames only around valid access events, minimizing storage and bandwidth.

The fusion result is published to EPICS as standardized process variables following a facility:subsystem:function:attribute naming convention across Access, Vision, Fusion, and Interlock subsystems. Alarm severity fields follow the EPICS alarm model, and the interlock output is a fault-safe enum record wired to a hardwired relay: loss of health defaults to "block," requiring explicit human reset. The system is deployed across two network segments via an edge gateway.

We describe the architecture, PV design, and state-machine logic, and report the evaluation framework — detection rate, false-positive rate, and end-to-end latency — comparing access-only, vision-only, and fused approaches. Because all underlying components are standard (OPC UA, a commercial detector, and EPICS), the architecture serves as a reusable template for other EPICS-based accelerator facilities.

Speaker: huliang hu (北京大学)

53 Tail-Latency Characterization of EPICS CA and PVA for Multi-Camera Diagnostics in Petawatt Laser Facilities

High-repetition-rate petawatt laser facilities require deterministic and stable large-scale image transfer for online multi-camera diagnostics. Even when cameras and IOCs maintain steady frame updates, complete image-group readout on the client side may experience severe tail latency, introducing uncertainty into online visualization, data logging, and feedback control. In this work, we establish a GigE Vision multi-camera testbed representative of the diagnostic scale of the CLAPA-II petawatt laser proton accelerator and systematically evaluate the end-to-end performance of EPICS Channel Access (CA) and PV Access (PVA) under synchronous serial client readout.

Under 10 GbE networking with ten cameras operating at 20 fps, both CA and PVA achieve an aggregate image production rate of approximately 466 MiB/s on the IOC side. However, significant differences emerge in client-side complete-cycle readout latency. For CA, maximum readout times range from 30.97 to 58.34 s, whereas PVA remains within 1.79 s. Under a bandwidth-constrained 1 GbE configuration with four cameras, the 95th-percentile (p95) readout latency reaches 67.13 s for CA, compared with 3.61 s for PVA. Packet-level tracing reveals that the long-tail latency observed with CA is accompanied by persistent TCP retransmissions, zero-window events, and window-full conditions, indicating complex interactions between serial request scheduling and transport-layer backpressure.

Based on these observations, we propose a production-consumer rate separation evaluation framework, emphasizing that control-system acceptance criteria should consider not only average throughput but also tail-latency metrics, including p95/p99 latency and maximum complete-group readout time. Finally, we discuss practical deployment considerations for EPICS areaDetector-based multi-camera diagnostic systems.

Speaker: 梦凡 曹 (北京大学物理学院)

54 Development and Performance of In-House Femtosecond Synchronization Control Hardware for S³FEL

Femtosecond synchronization is essential for the coordinated operation of high-repetition-rate free-electron lasers and ultrafast experiments. To meet the synchronization requirements of the S³FEL project, we have designed and developed all synchronization control hardware in-house, enabling customization for different synchronization tasks and supporting long-term maintenance and upgrades. This talk will briefly introduce the motivation for in-house development, present the developed hardware, and summarize recent synchronization test results. An RMS timing jitter of 10 fs, integrated from 10 Hz to 10 MHz, has been achieved using the in-house hardware.

Speaker: 俊淞 丁 (深圳先进光源研究院)

55 Design and Application of the HEPS Beamline Motion Control System (MCS2)

The High Energy Photon Source (HEPS), China's first fourth-generation synchrotron radiation facility, is designed to provide extremely high brightness and coherence for cutting-edge scientific research. Its Phase I project encompasses 15 beamlines, requiring precise motion control for approximately 1800 axes of optical components and endstation devices. To address the stringent demands of sub-micron positioning accuracy, multi-axis synchronous motion, and real-time data acquisition under constrained budget, we designed and implemented the HEPS Beamline Motion Control System (MCS2). The system adopts a distributed hierarchical architecture: at the hardware layer, it is based on EtherCAT fieldbus, employing ACS programmable controllers, self-developed driver cards, and integrated boards; at the software layer, it is built on EPICS with a multi-IOC architecture (Motor IOC and Script & Data Acquisition IOC), providing basic motion control, real-time script execution (trajectory planning, multi-axis coordination, triggering), and multi-channel data acquisition up to 20 kHz. This report presents the design approach, hardware implementation, and software functionality of MCS2, along with its application in general-purpose stepper axis motion control for HEPS beamlines.

Speaker: Dianshuai Zhang (高能所)

56 Operation and Improvement of the Fast Machine Protection System for HEPS

The fast protection system (FPS) of the High Energy Photon Source (HEPS) is a key interlock system for ensuring safe accelerator operation. Since being put into operation, the system has operated stably and reliably. It has triggered interlock protection multiple times in actual operation, verifying the reliability of its interlock logic and protection actions. During operation, interlock conditions have been flexibly configured and adjusted according to actual operational requirements, providing high system adaptability while ensuring protection functions. To support beam experiments, an adjustable trigger interval function was added. In addition, a signal interface with the timing system was implemented; through the timing system, trigger signals are distributed to all BPMs around the storage ring to meet experimental data acquisition requirements. Meanwhile, the operator interface (OPI) was optimized, making experimental operation and status monitoring more concise and intuitive. Operational results show that the system ensures safe operation of the HEPS accelerator and effectively supports beam commissioning and beam experiments, with good reliability, flexibility, and maintainability.

Speaker: zh ma

17:55 赞助商深信服

Wednesday, 4 November

System Management

57 Development of EPICS Control System for Laser Proton Accelerators and Radiotherapy Machines

A laser-driven proton accelerator based on the 2 PW ultra-intense laser facility is under development at Peking University, aiming at compact proton radiotherapy applications[1]. Compared with conventional RF accelerators, laser-driven acceleration features shot-to-shot operation, broad energy spectra, and large parameter fluctuations, which impose new requirements on the control system, including high-repetition-rate data acquisition for the laser facility, timing control for laser-target shooting and beamline transmutation[2], shot-based data archiving, and closed-loop coordination among the laser, target chamber, beamline, and patient treatment workflow.
To address these challenges, we are developing an EPICS-based control system for the laser proton accelerator and the associated radiotherapy machine. In this contribution, we present the overall architecture of the control system, including the IOC design for the petawatt laser control, beam transport and target chamber control, and proton beamline transport control; the interface design between the commercial treatment control software and the laser-driven accelerator control system; and the data management strategy for shot-based operation. We also report the development of operator interfaces (Phoebus), the personnel safety interlock system[3], the machine protection system, and the state management system integration, together with the commissioning and overall test results of the facility. The progress, experience, and future plan towards a clinical-grade EPICS control system for laser-based radiotherapy will be discussed[4].
Furthermore, we have preliminarily explored large language model applications built upon the Osprey framework [5] to enable seamless integration with the EPICS architecture, realizing intelligent perception and autonomous decision-making for the laser proton accelerator and building its dedicated intelligent agent.
Reference
[1] K.D. Wang, Y.Y. Zhao, K. Zhu, et al., "Review of the research progress of laser accelerator CLAPA-II," Sci. Sin.-Phys. Mech. Astron. 53, 100009 (2023). doi: 10.1360/SSPMA-2022-0433
[2] Q.Y. He, Q. Wang, K. Chen, et al., "Design and implementation of the CLAPA-II timing system," JINST 20, P10032 (2025). doi: 10.1088/1748-0221/20/10/P10032
[3] M.X. Zang, Q. Wang, E.S. Guo, et al., "Design and development of the personnel protection system for a petawatt laser accelerator" (2026), accepted for publication.
[4] K. Chen, K.D. Wang, M.F. Cao, et al., "A model-based adaptive dose delivery method for laser-accelerated proton beams," Mach. Learn.: Sci. Technol. 7, 045055 (2026).
[5] T. Hellert, J. Montenegro, and A. Sulc, "Osprey: Production-Ready Agentic AI for Safety-Critical Control Systems," arXiv preprint arXiv:2508.15066 (2025).

Speaker: Mr 强 王 (Peking University)

58 基于EPICS的核孔膜辐照终端控制系统研究与设计

针对核孔膜制备过程中对重离子束流辐照均匀性、卷膜传输高精度同步性以及系统长期运行可靠性的严苛要求，本文设计并实现了一套基于实验物理与工业控制系统（EPICS）架构的分布式核孔膜辐照终端控制系统。该系统旨在解决多物理参量强耦合、控制逻辑复杂及实时性要求高的技术难题，实现了对辐照过程的全局闭环调控与精细化管理。在系统架构层面，采用分层式设计，以输入输出控制器（IOC）为核心，通过以太网及现场总线将卷膜运动控制、束流探测与反馈、真空及环境监测等子系统集成为统一的实时控制网络。针对卷膜系统的精密控制难题，建立了卷径实时估算模型与张力-速度双闭环控制算法，有效抑制了卷绕过程中的线速度波动与膜材褶皱，实现了恒张力、恒线速度的高品质传输。在束流探测与反馈控制方面，设计了多传感器探测系统，开发了束流强度、均匀度与注量的多参数控制策略，确保了核孔密度的均一性。系统集成了基于Channel Archiver的高效数据归档引擎，实现了对束流参数、运动状态、真空度等关键运行数据的秒级长周期可靠存储与历史追溯。同时，构建了基于专家知识与状态机模型的层次化故障诊断体系，可对传动过载、束流丢失、真空突变等异常事件进行毫秒级识别、定位与联锁保护。测试与试运行结果表明，该控制系统运行稳定，响应迅速，维护便捷，显著提升了核孔膜辐照的成品率与工艺重复性，为高附加值核孔膜的工业化连续生产提供了坚实的自动化技术支撑与工程实践参考。

Speaker: Dr li yunjie (Institute of Modern Physics, Chinese Academy of Sciences)

59 激光控制系统与智能调控

当前，先进阿秒激光设施正处于关键建设阶段，高能量、高重频飞秒激光作为其核心驱动源，对装置的整体性能具有决定性影响。为打破该领域关键技术长期依赖进口的局面，团队自2023年起正式启动超快激光核心技术的自主研发攻关，并于2025年底顺利完成激光研制项目的全面验收。本报告将重点介绍自主研发的激光控制系统，详细阐述其软件架构与数据体系设计、与EPICS控制系统的集成对接方案，以及在智能调控技术方面取得的阶段性进展；此外，还将对装置的总控系统作简要介绍。

Speaker: 吴 小璇

60 Systematic simplification of IOC development and deployment at IHEP

With our refactoring of the IOCs for Xspress3 as an example, the idea of systematic simplification of IOC development and deployment is introduced. With the ihep-pkg packaging system (including mingw-build) and the ~/iocBoot convention (plus ~/epics), the use of EPICS can be highly simplified on the db/cmd and upper levels. ADGenICam is extended to support the much wider range of devices with GenICam-like interfaces, greatly simplifying the C++ code of their IOCs. QueueIOC, our Python IOC framework that attempts to further simplify development, is presented with sequencers and monochromator IOCs as the main examples. Our Python IOCs for detectors are based on friendly, reusable simplified "SDKs", which may also be used in areaDetector-based IOCs. Also briefly discussed is our perspective on the use of techniques like AI and virtualisation/containerisation in the EPICS ecosystem.

Speaker: Yu Liu

61 IOC Dynamical Control In CSNS

An IOC Dynamical control tool was developed in CSNS. With this tool, an IOC agent is running as a companian with the running IOC in each machine. The IOC is runing with a IOCStats module, and the IOC agent collects the IOC running information provided by the IOCAdminSoft.db, and then send these informstion to a MQQT broker with MQTT protocol.A backend server consumes the MQTT messages and communicates with frontend with Webspket protocol.
The IOC is running as a service. The IOC agent can receive the start/stop/restart command from the frontend throw backend service, and starts/stops/restarts the corresponding IOC.
The frontend can simply deploy the EPICS enviroment (including base,synApps,caPutLog, epics-62541,IocStats, s7nodave, recCaster and IOC) to an empty virtual machine. The backend service then will discover the new IOC/IOC-Agent, and send its information to the frontend.

Speaker: UNKNOWN LI Mingtao (高能所)

62 Design and Implementation of a Kubernetes-Based Automated Operations and Maintenance System for a Laser Accelerator Control System

Large scientific facilities such as accelerators rely on multiple EPICS IOCs to coordinate experimental devices. However, distributed deployment and heterogeneous runtime environments make software updates, status monitoring, and fault handling difficult to manage consistently. To address the isolated network environment, complex software dependencies, and hardware-specific placement requirements at CLAPA-II, this work develops an automated operations platform for the control system based on Docker and Kubernetes. An offline Kubernetes cluster was deployed across 14 servers. A common EPICS base image and IOC-specific images are used to encapsulate the required runtime environments, while host networking and node affinity support EPICS communication and device connectivity. A web interface provides status monitoring, IOC start/stop control, and log access. For log management, caPutLog is integrated into the IOC images to record process variable (PV) write operations and the associated value changes. IOC runtime output is redirected via YAML deployment manifests and then collected by Filebeat for centralized storage; this supports fault diagnosis and the automatic generation and distribution of daily experiment records. A manual on-site restart typically takes more than five minutes; in node shutdown tests using an RCF motor IOC, PV access was automatically restored in approximately 53 s on average. The platform now supports all CLAPA-II IOCs and has been operating stably on site for six months. This work provides a practical approach to IOC containerization, automatic recovery, and centralized operations.

Speaker: 硕 金 (北京大学)

63 SHINE综合报警的设计与研发

上海硬X射线自由电子激光装置（SHINE）运行过程中产生大量设备状态、束流参数及报警信息，传统单点报警方式难以有效反映多设备、多系统之间的关联关系。针对这一问题，构建面向SHINE运行的综合报警与智能诊断系统。在底层保持EPICS CA/PVA设备接入方式的基础上，上层引入NNG通信框架，建立统一的数据交互机制，实现实时数据、Kafka报警信息及Archiver历史数据的统一接入与组织。系统结合PV语义、设备系统、物理区域、报警时间关系及历史数据变化，对离散报警进行事件聚合、相关PV筛选和关联分析，并形成异常事件及根因候选结果，为SHINE运行状态分析、异常定位和故障诊断提供辅助。

Speaker: 宇飞 张 (中国科学院上海高等研究院)

10:20 Coffee Break

64 Design of a Synchrotron Beamline Control System Based on EPICS and MicroTCA.4

Stable and reproducible synchrotron beamline operation requires coordinated control of optical elements, beam diagnostics, and detector acquisition. Integrating these functions with model-assisted optimization requires consistent interfaces between hardware execution, experimental data, and operating decisions. To address these requirements, a beamline control system based on EPICS and MicroTCA.4 is proposed. The design combines modular motion control, low-current measurement, high-speed acquisition, and event timing with an EPICS/synApps software layer for device access and experiment coordination. Motor and detector interfaces are integrated with Bluesky/Mamba workflows to associate measurements with actual positions, exposure windows, and calibration states during step and continuous scans. Physics-based optical models and data-driven surrogates provide the basis for predicting beam position and flux responses, with measurement-based calibration and independent evaluation included in the development strategy. Model-predictive optimization is coupled with local feedback through bounded setpoint updates, command validation, and independent PLC interlocks. The architecture also provides interfaces for digital-twin applications. A staged validation strategy links subsystem tests to system-level objectives for synchronization, data completeness, beam stability, and tuning efficiency. The resulting design establishes an explicit connection between hardware capabilities, EPICS software functions, and model-assisted control, providing an engineering basis for coordinated and intelligent beamline operation.

Speaker: Dr Yi Zhou (Nanjing University)

65 A Modular EPICS-Centered Data Acquisition Platform at DCLS and DALS

We developed a modular data acquisition platform that began at the Dalian Coherent Light Source (DCLS) and was extended in the current deployment at the Dalian Advanced Light Source (DALS). Its principal path archives operational data through EPICS Channel Access. PV subscriptions are distributed across independently supervised collector processes, allowing acquisition capacity to grow by adding processes. The PV list is periodically compared with an IOC management database; newly registered PVs are incorporated through configuration updates and controlled collector restarts. This provides an automated way to maintain coverage as the control system evolves.

As an alternative architecture for EPICS archiving, the platform stores typed values, source timestamps, and alarm information in TDengine. The time-series database also provides a common interface for monitoring, visualization, retrieval, export, and further analysis, so new services can be developed independently of the collectors. A separate high-rate path demonstrates the extensibility of this design: on a standard server, its software receiver has been tested with all 18 DALS beam position monitors operating at approximately 1 MHz per device. The presentation will discuss the implementation, operational experience, and measured capabilities of the EPICS path, and show how the shared data layer accommodates both routine operational signals and high-rate diagnostics.

Speaker: 惟彬 郑 (深圳先进光源研究院)

66 Study on Operation Upgrade and Optimization of the Accelerator Control System of Shanghai Synchrotron Radiation Facility

As a high-performance synchrotron radiation facility, the Shanghai Synchrotron Radiation Facility (SSRF) relies on its control system as the core support to ensure stable beam current, equipment safety and efficient photon delivery. To address issues encountered by the control system during long-term operation, including hardware aging, communication bottlenecks, end-of-support operating systems, and reliance on imported key components, systematic upgrading, renovation and operation & maintenance optimization have been carried out.

The traditional VME-based power supply control architecture has been replaced with SoftIOC and an Ethernet distributed architecture. Additional improvements include upgrading the insertion device feedforward control system, renovating control network switches, establishing a high-precision time-stamp system and historical data platform, developing domestic processor-based IOC controllers, and adopting a homegrown Linux operating system. These measures have comprehensively improved system stability and the level of independent controllability.

A comparison of operational data from 2022 and 2025 shows that the annual number of control system failures dropped from 29 to 8, beam downtime reduced from 4.29 hours to 1 hour, and the MTBF (Mean Time Between Failures) increased from 292 hours to 973 hours, representing a marked improvement in operational stability. This engineering practice can serve as a reference for the long-term operation, technical upgrading and localization substitution of control systems for synchrotron radiation light sources and similar large scientific facilities.

Speaker: 鹤云 王 (中国科学院上海高等研究院)

67 Establishment of the Operational Environment for the SHINE Accelerator Control System

This report presents the establishment of the operational environment for the SHINE accelerator control system. The environment is built on a virtualization infrastructure combining Sangfor virtual system with Anolis OS, providing autonomous and controllable capability as well as operational stability. An LDAP-based real-name account management system centralizes user identity storage and authentication, ensuring that every user accesses the environment with verified credentials. On this basis, channel-based permission settings implement fine-grained access control following a "role–resource–operation" mapping model, so that users in different roles—physicists, commissioning staff, engineering designers, and system administrators—obtain differentiated access channels according to their responsibilities. Standardized IOC templates further provide reusable software and configuration frameworks for the control layer, covering device communication, data processing, and security management. Together, these components form a secure, traceable, and extensible operational environment that provides stable and reliable support for the software operation of the SHINE accelerator control system.

Speaker: 清茹 米 (中科院上海高等研究院)

68 Design and Commissioning of a System for Beamline Equipment Operational Monitoring in HEPS

Purpose
To address the operational requirements of the High Energy Photon Source (HEPS) —
China's first fourth-generation synchrotron radiation facility — which demands multi-beamline support and high reliability, we designed and implemented a comprehensive
beamline equipment operational monitoring system. The system is designed not only
for real-time monitoring and anomaly alerting but also to support future predictive
maintenance, thereby ensuring the stable and efficient operation of the facility.
Methods
Built on the EPICS architecture, the system is structured from bottom to top into the
PLC layer, server layer, and user interface layer. It integrates monitoring functions from
the front end to the endstations, including vacuum monitoring, cooling monitoring,
equipment status monitoring, and the status monitoring of safety shutters and photon
shutters. The system enables real-time monitoring, alarm management, and historical
data recording. Through operational reports, it performs statistical analysis on key
metrics such as the vacuum and cooling alarm records, shutter switching frequency,
and fast gauge trigger counts.
Results
This system has been deployed on-site at HEPS, fully addressing the security
requirements of complex operational environments. Through a closed-loop process
covering data acquisition, analytical diagnosis, and preventive maintenance, it
effectively supports stable monitoring during the operation phase.

Speaker: Mr Jiahao Liu

69 PV Management and High-Performance Data Access System Based on EPICS Archiver Appliance

Within the EPICS framework, Archiver Appliance is commonly employed as the management and storage engine for Process Variable (PV) data. Its stability and ease of use have been validated over the long term at major international scientific facilities such as LCLS, NSLS-II, and LANSCE. However, as the scale of control systems continues to expand—the accelerator section of the China Initiative Accelerator Driven System (CiADS) currently has 232,070 PVs under archival, a number expected to exceed 300,000—PV registration, archival status tracking, and anomaly troubleshooting can no longer rely solely on native functionalities. Existing storage approaches are also becoming bottlenecks in terms of the validity, security, and access efficiency of massive historical data.

To address these challenges, we have carried out development work on two fronts based on Archiver Appliance. First, we established a unified PV information management and operational inspection platform, bringing hundreds of thousands of PVs and their archival statuses into a controllable, traceable management workflow. Second, we introduced a time-series database designed for long-term storage and high-performance retrieval, enabling rapid search of massive archival data and direct use in subsequent analyses. The former operates as a web service interfacing with the native Archiver Appliance APIs, enabling unified management of primary/backup redundant archival systems, batch submission and audit trails for PV archival requests, classification and retrieval of PVs by subsystem and segment, as well as monitoring views for various PV anomaly states with scheduled inspections and automatic recovery. The latter adds a new storage interface class within Archiver Appliance to write data in real time to the domestic time-series database DolphinDB. A stream data processor then categorizes PVs into low-frequency, medium-to-high-frequency, and waveform types, establishing separate databases for each. Storage and indexing are optimized according to the respective frequency and data-type characteristics, achieving cold/hot data separation and periodic cleanup of invalid data. This significantly reduces data retrieval time for massive datasets, compresses storage footprint, and provides users with standardized data processing capabilities—such as timestamp alignment—through built-in database functions.

Keywords: EPICS; Archiver Appliance; DolphinDB; PV information management; time-series database; stream data processing

Speaker: 钊扬 李 (中国科学院近代物理研究所)

Lunch

Visit CSNS facility

Banquet

Thursday, 5 November

Training session

70 MTCA

Speakers: Gabriel Fedel (European Spallation Source), Joao Paulo Martins (European Spallation Source ERIC)

10:20 Coffee Break

71 EPICS Driver

Speaker: Kazuro Furukawa

Training session 2

72 Phoebus

Speaker: Sky Brewer (ESS)

10:20 Coffe Break

73 Archiver

Speaker: Sky Brewer (ESS)

Lunch

Training session

74 Timing

Speakers: Gabriel Fedel (European Spallation Source), Joao Paulo Martins (European Spallation Source ERIC)

15:20 Coffee Break

75 Timing

Speaker: Ming Liu

Training session 2

76 OPC UA (Online)

Speaker: Ralph Lange

15:20 Coffee Break

77 OAC Tree Sequencer (Online)

Speaker: Walter Van Herck

Friday, 6 November

Training session

78 EPICS 7

Speaker: Simon Rose (European Spallation Source ERIC)

10:20 Coffee Break

79 Stream Device

Speaker: Simon Rose (European Spallation Source ERIC)

Training session 2

80 areaDetector (Online)

Speaker: Érico Nogueira Rolim

10:20 Coffee Break

81 Pyapas

Speaker: 晓含（Xiaohan） 卢 (Lu) (高能所)

Lunch