Abstract: To address the challenges of accessing massive data and making reasonable use of information and communication infrastructure resources, this paper discusses the key technologies of the cloud-edge-end collaborative system in smart rail transit. The overall framework and technical architecture of smart rail transit systems are summarized, emphasizing and the common characteristics of the intelligent evolution of each professional system are presented. The management, processing capabilities, bandwidth occupancy, real-time and other characteristics of cloud centers, edge computing platforms and terminals are analyzed, cloud-edge collaboration methods for smart rail transit are studied, and matching system hierarchical architecture and network architecture are proposed.  On this basis, taking the catenary intelligent maintenance system as an example, a typical system solution is proposed in which a single maintenance vehicle is equipped with edge computing nodes to integrate various intelligent terminals to achieve independent detection and maintenance capabilities, and is managed, optimized, and iterated through a cloud center. The cloud-edge-end collaborative system helps achieve effective integration between ubiquitous sensing , efficient processing, and information and communication infrastructure in rail transit, presenting broad prospects in smart rail transit.

Key words: smart rail transit, cloud-edge-end, intelligent maintenance, edge intelligence, system architecture

摘要： 为解决智慧轨道交通中海量数据接入的同时合理利用信息和通信基础设施资源，研究智慧轨道交通云边端协同系统的关键技术。归纳智慧轨道交通系统的总体框架与技术架构，提出各专业系统智能化演进的共性特征。分析云中心、边缘计算平台及终端的管理、处理能力、带宽占用、实时性等特点，研究智慧轨道交通云边端协同方式，并提出相匹配的系统分级架构与网络架构。在此基础上，以接触网智能运维系统为例，提出单台检修车设置边缘计算节点集成各类智能终端实现独立检测维修能力，通过云中心统一管理、优化、迭代的典型系统方案。云边端协同有助于实现轨道交通泛在感知、高效处理与信息、通信基础设施之间的有效融合，在智慧轨道交通中有广泛的应用前景。

关键词: 智慧轨道交通, 云边端, 智能维修, 边缘智能, 系统架构