【摘要】 针对大规模星座卫星数目多、管控数据量大，利用传统人工和图表方式管理星座数据已经无法满足任务时效性等要求，提出基于自然语言的星座人智协同管控方法，以航天专业知识和管控指令库为指导，解析并执行用户自然语言输入的管控指令。建立了通用的管控语句模型和通用指令模型，提出基于自然语言的人智协同管控架构，使用户可以通过模糊自然语言查询星座信息与发送指令。该方法可以避免人工管理层层解析的过程，自动抽象用户管控语句特征，提高准确率与查询效率。设计了智能决策与反馈机制，为未来利用大模型实现星座管控提出一种解决思路，通过支持集成不同的算法库实现决策序列的智能推荐，为用户提供行动选项，以此优化决策流程并提高管控效率。仿真案例表明，所建立通用的管控语句模型和通用指令模型对遥感星座具有准确的信息抓取能力和指控能力。更多还原

【Abstract】 Faced with the challenges posed by large-scale satellite constellations, such as the substantial number of satellites and the overwhelming volume of management data, traditional manual and chart-based approaches are deemed inadequate for meeting mission timeliness requirements. To mitigate these issues, a natural language-based human-AI collaborative control methodology is introduced. This approach, guided by aerospace expertise and a repository of control commands, parses and executes user-inputted natural language control instructions. A universal control statement model and a generalized instruction model are formulated, and a natural language-driven human-AI collaborative control framework is proposed, allowing users to retrieve constellation information and issue commands using ambiguous natural language queries. This methodology circumvents the need for multi-layered parsing in manual management, automatically abstracts key features from user control statements, and improves both accuracy and query efficiency. Additionally, an intelligent decision-making and feedback mechanism is devised, offering a solution for future constellation control leveraging large-scale language models. By supporting the integration of diverse algorithm libraries, the mechanism facilitates intelligent recommendation of decision sequences, providing users with actionable options to streamline the decision-making process and enhance control efficiency. Simulation results confirm that the established universal control statement model and generalized instruction model demonstrate precise information retrieval and effective command execution capabilities for remote sensing constellations.更多还原