用于EPICS输入输出控制器的冗余技术研究

Availability is an important parameter of the large-scale scientific facilities, it receives more and more attention with the recent rapid development of user facilities, such as synchrotron radiation light source. Availability is paid great attention to in the projects related to nuclear energy, such as C-ADS(Chinese ADS) and TMSR(Thorium Molten Salt Reactor). EPICS is currently the most widely used control system development platform for the large-scale scientific facilities all over the world. It has been applied successfully in nearly one hundred large-scale scientific facilities, involving large particle accelerators, high-energy physics detectors, telescope，etc. By analyzing the components of the EPICS based control system, the input-output controller is the only component, for which there are not mature redundancy solutions, so it has become the single point of failure in the entire control system based on EPICS. In order to solve this problem, a new redundancy solution is proposed. The research will cover the redundancy structure, resource management and device drivers of EPICS input-output controller, a prototype system will be setup to check the feasibility of the redundancy solution. The project proposal can improve the availability of the EPICS based control system, and has very important value in engineering application.

可用性（availability）是大型科学装置的一个重要指标，随着近年来同步辐射光源等用户装置的快速发展而受到了越来越多的关注，在国内启动的C-ADS和TMSR等核能项目中可用性更是受到了高度重视。EPICS是目前国际上应用范围最广的大型科学装置控制系统开发平台，已成功应用于全世界近百个大型科学装置，涉及大型粒子加速器、高能物理探测器和天文望远镜等多种类型装置。通过对EPICS控制系统组成部分的分析，发现输入输出控制器是目前唯一还没有成熟冗余解决方案的部分，使得它成了EPICS控制系统中的单点故障（single point of failure）。本项目针对这一问题，提出一套新的冗余技术方案，开展EPICS输入输出控制器的冗余结构、资源管理和设备驱动等方面的研究，并建立一套原型系统来检验冗余技术方案的可行性。本项目的开展，可以提高EPICS控制系统的可用性，具有非常重大的工程应用价值。

本项目围绕EPICS IOC的冗余性问题，对PI (PROFIBUS & PROFINET International)发布的“PROFINET IO System Redundancy”冗余技术标准进行了研究，明确了分层冗余设计的方法，每层的冗余配置保持独立，与其它层无关。对目前市场上存在的冗余技术产品进行了广泛调研，设计了一个基于虚拟化技术和PROFINET的全冗余EPICS控制系统解决方案。该方案将商业化产品引入到EPICS环境中，具有可行性高、易于实现的特点。开发出一套原型系统，并对其冗余性能进行了测试，得到了各层的冗余切换时间，EPICS IOC的冗余切换时间约为800ms，PLC的冗余切换时间约为6ms，MRP环的冗余切换时间约为70ms。这些数据对于将原型系统应用于工程实际具有重要的指导意义。根据“基于可调谐红外激光的能源化学研究大型实验装置”（FELiChEM）硬件联锁保护系统的性能要求，我们将本项目提出的全冗余EPICS控制系统解决方案应用于FELiChEM硬件联锁保护系统的设计，以充分验证该冗余解决方案的工程应用价值。