近年来，中国城市轨道交通建设取得了令人瞩目的成就。与此同时，城市轨道交通运维企业也面临着安全运营、高效运营的挑战：在资产管理方面，截至2023年年底，我国城市轨道交通全行业平均资产负债率为57%，普遍面临经营亏损和人员优化的难题；在运营管理方面，疫情之后随着客流持续攀升，运营组织持续承压，设施设备科学维保已被提升到了前所未有的高度；在技术管理方面，智能运维是大势所趋，但运维侧科技创新能力不足，难以有效支撑和落实城市轨道交通高质量发展的目标。

因此，未来5至10年内城市轨道交通行业最具挑战的任务将是在确保企业财务健康的前提下,利用新质生产力处理好设施设备不断老化与运营管理安全高效之间的矛盾。在这样的背景下，AI技术的快速发展将给行业带来新的解题思路。

AI能助力城市轨道交通资产全寿命周期管理。车辆、通号、供电、工务等专业系统的运维管理涉及规划、采购、运营、维修、退役等多个管控环节，各环节都会产生大量的设备状态数据、故障数据和生产管理数据。当下越来越多的管理者开始意识到资产全寿命周期管理对成本管理和生产质量控制的重要性。AI技术的演进，尤其是DeepSeek等先进算法的突破，为城市轨道交通资产全寿命周期管理提供了更多可能。在传统检修生产模式中，设备生产数据基本上都是靠人工输入，由于作业流程标准化程度不高，数据质量参差不齐，极大地影响了城市轨道交通运维企业数字化转型的成效。如果能实现数据自动化采集，再将大量高质量运维数据自动投喂给算法模型进行训练，那么“数据驱动资产全寿命周期管理决策”将成为可能。因此，在AI算力、算法和数据三大要素中，城市轨道交通设施设备高质量运维数据自动化采集对于智能运维尤为重要。

在管理实践中，人们已经看到AI赋能城市轨道交通行业在多个领域展现出巨大价值。其中最具代表性的是AI具身智能，即智能体通过与身体相关的感知和行动来实现的智能，可助力实现城市轨道交通运维企业智能化管理闭环运行，实现资产全寿命周期中经济效益和社会效益的平衡。AI技术领先的公司，能为城市轨道交通运维企业提供三类解决方案：一是感知数智化解决方案，即在具身智能领域，能够为运维企业提供高精确度、低误报、作业高效的基地智检机器人+轨旁360°检测、区间线性资产设备检测、车站设备检测等全场景智能运维解决方案；二是应用数智化解决方案，即AI公司可助力将运维企业既有BI（Business Intelligence，即商业智能工具）可视化平台和EAM（Enterprise Asset Management，即资产管理系统）与感知侧AI智能数据分析平台高度集成，全智能处理设施设备日常运维数据“汇、存、管、算、用”的整个过程，最大程度降低人工参与数据的生成，确保数据质量，为AI模型训练和AI能力调用提供基础保障； 三是生产数智化解决方案,这类AI公司能将铁路领域关键系统部件修自动化生产技术赋能给城市轨道交通车辆架大修生产，帮助运维企业解决日益突出的车辆架大修和部件修工艺技能短板和生产能力问题。

预测轨道交通AI应用的未来前景，必将实现全链条数智化与生态协同。可以相信在产业生态圈的共同推动下，“城轨AI+应用”将进一步深化，最终实现资产全寿命周期端到端数智化协同。主要发展方向包括以下三方面：一是AI具身智能+5G，具备行业知识和技能的“数字员工”能够专业地胜任运维作业；二是AI+BI+数字孪生，构建全息城市轨道交通设施设备运维管理智能体；三是AI+跨专业数据时空标签同步+行业知识图谱，实现车辆与供电、车辆与工务等跨专业智能运维协同，提高故障诊断能力。

AI赋能城市轨道交通高质量发展，已经从概念探索进入实际落地阶段。从自动感知到自主决策，AI技术正在深刻改变城市轨道交通行业的生产流程、作业模式和装备升级。展望未来，中国城市轨道交通行业必将在全球率先实现AI驱动的智能运维体系，为全球智能交通发展提供“中国解决方案”。

In recent years, urban rail transit construction in China has achieved remarkable success. Meanwhile, urban rail transit operation and maintenance enterprises are also facing challenges in ensuring safe and efficient operations. In terms of asset management, by the end of 2023, the average asset-liability ratio of China′s urban rail transit industry was 57%, and most enterprises generally faced the problems of operating losses and personnel optimization. In terms of operation management, after the COVID-19 pandemic, with the continuous increase in passenger flow, the pressure on operation organization has been continuously rising, and the scientific maintenance of facilities and equipment has been elevated to an unprecedented level. In terms of technical management, intelligent operation and maintenance is the general trend, but the lack of scientific and technological innovation capabilities on the operation and maintenance makes it difficult to effectively support and implement the goal of high-quality development of urban rail transit.Therefore, in the next 5 to 10 years, the most challenging task for the urban rail transit industry will be to handle the contradiction between the continuous aging of facilities and equipment and the safe and efficient operation management, while ensuring the financial health of enterprises, by leveraging new forms of productive forces. Against this backdrop, the rapid development of AI technology will bring new solutions to the industry.

AI can assist in the whole life cycle management of urban rail transit assets. The operation and maintenance management of technical subsystems such as vehicles, communication and signaling, power supply, and track works involves multiple management and control links including planning, procurement, operation, maintenance, and decommissioning. Each link generates a large amount of equipment status data, fault data, and production management data. Nowadays, more and more managers are beginning to realize the significance of the whole life cycle management of assets for cost management and production quality control. The evolution of AI technology, especially the breakthroughs in advanced algorithms such as DeepSeek, provides more possibilities for the whole life cycle management of urban rail transit assets. In the traditional maintenance and production mode, equipment production data is basically input manually. Due to the low standardization of the operation process, the data quality varies greatly, which greatly affects the effectiveness of the digital transformation of urban rail transit operation and maintenance enterprises. If automated data collection can be achieved, and a large amount of high-quality operation and maintenance data can be automatically fed into the algorithm model for training, then "data-driven decision-making for the whole life cycle management of assets" will become possible. Therefore, among the three key elements of AI computing power, algorithms, and data, the automated collection of high-quality operation and maintenance data of urban rail transit facilities and equipment is particularly important for intelligent operation and maintenance.

In management practice, it has already been witnessed that AI empowering the urban rail transit industry demonstrates tremendous value in multiple fields. The most representative one is AI embodied intelligence, the intelligence of an intelligent agent realized through perception and actions related to its body. It can help achieve the closed-loop operation of intelligent management in urban rail transit operation and maintenance enterprises and balance the economic and social benefits in the whole life cycle of assets. Companies with leading AI technology can provide three types of solutions for urban rail transit operation and maintenance enterprises. The first is the sensing digital intelligence solution. In the field of embodied intelligence, it can provide operation and maintenance enterprises with full-scene intelligent operation and maintenance solutions such as base intelligent inspection robots with high precision, low false alarm rate, and high operation efficiency, 360° trackside detection, linear asset equipment detection in sections, and station equipment detection. The second is the application digital intelligence solution. AI companies can help highly integrate the existing BI (Business Intelligence) visualization platform and EAM (Enterprise Asset Management) system of operation and maintenance enterprises with the AI intelligent data analysis platform on the sensing side, and fully intelligently handle the entire process of "collecting, storing, managing, calculating, and using" the daily operation and maintenance data of facilities and equipment, minimizing the manual participation in data generation, ensuring data quality, and providing a basic guarantee for AI model training and AI capability invocation. The third is the production digital intelligence solution. Such AI companies can empower the production of major overhauls of urban rail transit vehicles with the automated production technology of key system components in the railway field, helping operation and maintenance enterprises solve the increasingly prominent problems of skills shortages and production capacity in the major overhauls of vehicles and component repairs.

It is expected that the future prospects of AI applications in urban rail transit will surely achieve full-chain digital intelligence and ecological synergy. It is believable that with the joint promotion of the industrial ecosystem, the "Urban Rail AI plus Application" will be further deepened to achieve end-to-end digital intelligence synergy in the whole life cycle of assets. The main development directions include the following three aspects: First, when AI embodied intelligence is combined with 5G, "digital employees" with industry knowledge and skills can professionally perform operation and maintenance tasks. Second, AI combined with BI and digital twin will build a holographic intelligent agent for the operation and maintenance management of urban rail transit facilities and equipment. Third, when AI is combined with the synchronization of cross-subsystem data spatio-temporal tags and an industry knowledge graph, it will achieve cross-subsystem intelligent operation and maintenance synergy between vehicles and power supply, vehicles and track works, etc., and improve the fault diagnosis ability.

AI empowering the high-quality development of urban rail transit has already entered the practical implementation stage from the conceptual exploration stage. From automatic sensing to autonomous decision-making, AI technology is profoundly changing the production processes, operation modes, and equipment upgrades in the urban rail transit industry. Looking ahead, the urban rail transit industry in China will surely take the lead in implementing an AI-driven intelligent operation and maintenance system globally, providing a "Chinese solution" for the development of global intelligent transportation.