一、方向简介
民用飞机生产安全研究面向航空制造全过程中的人员、设备、工艺、环境与质量协同安全问题，聚焦生产现场风险识别、工序质量控制、设备状态监测与作业行为规范等关键环节。该方向围绕装配作业安全、关键工艺过程监测、生产异常检测、工业视觉检测、人员不安全行为识别和制造流程优化等内容开展研究，综合运用人工智能、工业互联网、机器视觉、数字孪生与数据驱动分析技术，构建覆盖制造过程感知、风险预警、质量追溯与辅助决策的智能化体系。通过提升对复杂制造场景中隐患、偏差和异常的识别与处置能力，可有效支撑民用飞机生产过程的高可靠运行和高标准质量保障，为航空制造业的智能化升级、安全管理优化和高质量发展提供技术支持。
I. Introduction to the Direction
Civil aircraft production safety research addresses coordinated safety issues involving personnel, equipment, processes, environment, and quality throughout the entire aviation manufacturing process, focusing on key areas such as production site risk identification, process quality control, equipment condition monitoring, and standardized operational behavior. This direction conducts research on assembly operation safety, monitoring of critical manufacturing processes, production anomaly detection, industrial visual inspection, identification of unsafe personnel behaviors, and manufacturing process optimization. It comprehensively applies artificial intelligence, the Industrial Internet, machine vision, digital twins, and data-driven analysis technologies to build an intelligent system covering manufacturing process perception, risk warning, quality traceability, and decision-making support. By improving the ability to identify and handle potential hazards, deviations, and anomalies in complex manufacturing scenarios, the research can effectively support highly reliable operation and high-standard quality assurance throughout the civil aircraft production process, providing technical support for the intelligent upgrading, safety management optimization, and high-quality development of the aviation manufacturing industry.
二、研究内容与成果
VLM驱动的“人机环管”一体化平台
平台采用云计算技术，构建了包含“场景层、数采层、算法层、应用层的四层逻辑架构。平台的核心是安全大数据系统与视觉语言大模型（VLM），通过机器学习与可视化技术，为上层应用提供强大的AI分析与决策能力。围绕飞机制造中“人、机、环、管”四大要素 ，构建了“核心安全管控”和“绿色智能治理”两大应用集群。
II. Research Content and Achievements
VLM-Driven Integrated “Human-Machine-Environment-Management” Platform
The platform adopts cloud computing technology to build a four-layer logical architecture comprising the “scenario layer, data acquisition layer, algorithm layer, and application layer.” The core of the platform consists of a safety big data system and a Vision-Language Model (VLM), which provide powerful AI analysis and decision-making capabilities for upper-level applications through machine learning and visualization technologies. Focusing on the four key elements of “human, machine, environment, and management” in aircraft manufacturing, two major application clusters, namely “core safety management and control” and “green intelligent governance,” have been developed.
课题组通过VLM驱动的危险作业实时感知，突破了传统“人工巡检、事后响应”的被动模式，实现了对高危作业的实时智能感知，如“反光背心违规”“违规打电话”等行为的实时感知与自动报警。
Through VLM-driven real-time perception of hazardous operations, the research group has overcome the traditional passive mode of “manual inspection and post-event response,” enabling real-time intelligent perception of high-risk operations, such as the real-time detection and automatic alerting of behaviors including “non-compliant use of reflective vests” and “unauthorized phone calls.”
平台应用及推广
平台已在中国某飞机制造企业40余个部门规模化部署，形成了可复制的绿色安全制造新模式，推动企业管理从“经验驱动、分散治理”转向“数据驱动、模型预测”。成果实现了显著的经济、社会与生态效益，整体安全运营成本降低18%，并带动了AI算法、智能终端等新兴产业链聚集。其模块化、标准化的设计，为全球航空、轨道交通、船舶等高端装备制造业的绿色与智能化转型提供了可复制、可推广的“实践范例”。
Platform Application and Deployment
The platform has been deployed at scale across more than 40 departments of an aircraft manufacturing company in China, forming a replicable new model for green and safe manufacturing and promoting a shift in enterprise management from “experience-driven, decentralized governance” to “data-driven, model-based prediction.” The achievements have generated significant economic, social, and ecological benefits, reducing overall safety operation costs by 18% and contributing to the clustering of emerging industrial chains such as AI algorithms and intelligent terminals. Its modular and standardized design provides a replicable and scalable “practical model” for the green and intelligent transformation of high-end equipment manufacturing industries, including aviation, rail transportation, and shipbuilding, worldwide.
