上线投用的人工智能研发与应用全栈生产线基于国家能源集团自建的人工智能训练平台（研发侧），以及自主研发的智能视频分析系统（应用侧）构建而成，运用了包括皮带跑偏、大块煤识别、车辆长时停留、人员跌倒识别、除铁器异物识别、电子围栏、明火识别、安全帽识别等多达20多种算法模型、330多台监控摄像机24小时画面数据以及12台算法服务器的600多套算法路数。

该条生产线上线投用标志着在全集团范围内首次实现了工业化模式下人工智能技术从研发到应用层面的全域贯通，基于对人工智能数据与算法模型的高效复用，进一步助力工业场景下人工智能技术的优化升级，持续提高算法模型识别的准确率，为后续人工智能技术应用产业的高质量发展奠定了坚实基础。

目前，该条生产线在新疆公司及其下属乌东、准东、红沙泉、黑山四矿完成部署，构建起“二级公司-厂矿侧”煤炭生产二级监管体系，依托人工智能技术，实现了对实际生产环境的实时监控、智能分析，可多维度、多角度的实时监管设备运行状态和人员行为，发现异常自动弹屏报警，高效助力煤矿安全、高效生产。据了解，该条生产线在电力、化工等其他能源领域也具有广泛应用前景。

The artificial intelligence R&D and application full-stack production line is built based on the artificial intelligence training platform (R&D side) built by National Energy Group and the intelligent video analysis system (application side) independently developed. It uses more than 20 kinds of algorithm models, including belt deviation, identification of large coal, long-term vehicle stay, personnel fall identification, identification of foreign objects in iron removal, electronic fence, open flame identification, hard hat identification, more than 330 surveillance cameras 24 hours picture data and more than 600 sets of algorithms of 12 algorithm servers.

The launch of this production line marks the first time in the whole Group to realize the full penetration of artificial intelligence technology from research and development to application in the industrial mode. Based on the efficient reuse of artificial intelligence data and algorithm models, it further helps the optimization and upgrading of artificial intelligence technology in industrial scenarios, and continuously improves the accuracy of algorithm model recognition. It has laid a solid foundation for the high-quality development of the subsequent artificial intelligence technology application industry.

At present, the production line has been deployed in Xinjiang Company and its subordinate Wudong, Jundong, Hongshaquan and Heishan mines to build a "secondary company-factory side" coal production secondary supervision system. Relying on artificial intelligence technology, it realizes real-time monitoring and intelligent analysis of the actual production environment, and can monitor equipment operation status and personnel behavior in a multi-dimensional and multi-angle manner. Find abnormal automatic screen alarm, efficient help coal mine safety and efficient production. It is understood that the production line also has wide application prospects in other energy fields such as electric power and chemical industry.