|
谭娟(1975—),女,博士研究生,研究员,研究方向为遥感卫星应用、数据挖掘。 |
|
郭琦(1993—),男,硕士研究生,工程师。 |
Copy editor: 李楠
收稿日期: 2024-11-05
修回日期: 2024-12-09
网络出版日期: 2025-11-22
Space observation task derivation based on multidimensional data association
Received date: 2024-11-05
Revised date: 2024-12-09
Online published: 2025-11-22
随着各行业遥感观测需求越来越旺盛,当前航天观测资源“先用户申请,后被动响应”的应用模式的局限性越来越大。为此,开展基于多维数据关联的航天观测任务衍生技术研究。首先,通过对大规模差异化航天资源进行统一建模,实现对各类航天资源能力属性以及航天观测任务的准确描述。其次,针对各类显式或隐式航天观测需求,采用基于大语言模型方法,进行需求内容理解及任务要素抽取;然后,构建基于多元语义推理网络实现航天观测任务的衍生推理。并在此基础上,挖掘典型用户历史任务-资源供需关系变化规律,进行航天资源自动推荐。最后,以应急救灾场景中航天观测任务的衍生过程为例对本方法进行验证。实例分析表明,本方法在保证准确性的同时可有效提高航天观测系统的运行效率和使用效益。
谭娟 , 郭琦 , 张学亮 , 张文宝 . 基于多维数据关联的航天观测任务衍生技术研究[J]. 指挥控制与仿真, 2025 , 47(6) : 69 -75 . DOI: 10.3969/j.issn.1673-3819.2025.06.010
With the increasing demand for remote sensing observation from users in various industries, the current application mode of "passive response to user demand" for space observation resources is becoming more and more limited. Therefore, research on space observation mission derivation technology based on multidimensional data association is carried out. Firstly, the differentiated space resources are modeled to accurately describe the capability attributes of various space resources and space observation mission knowledge. Secondly, for all kinds of explicit or implicit space observation requirements, the method based on large language model is used to understand the requirements and extract the task elements; Then, a multi-semantic inference network is constructed to implement the derived reasoning of aerospace observation tasks. After that, the change rules of the historical task resource supply and demand relationship of typical users are mined to automatically recommend space resources. Finally, the derivation process of space observation mission in emergency disaster relief scenario is taken as an example to verify the method. The example analysis shows that this method can not only ensure the accuracy, but also effectively improve the operation efficiency and use efficiency of the aerospace observation system.
| [1] |
朱光熙, 王港, 张超, 等. 基于多模态观测需求信息的遥感星群任务智能规划机制[J]. 天地一体化信息网络, 2022(3): 23-29.
|
| [2] |
许宇栋, 周敬博, 尹嘉昭, 等. 对地观测卫星任务规划策略及应用研究综述[J]. 无线电工程, 2021, 51(8): 681-690.
|
| [3] |
张学飞, 张丽萍, 等. 知识图谱与大语言模型协同的个性化学习推荐[J/OL]. 计算机应用. 2024, 20241014.
|
| [4] |
陈恒, 马鑫君, 梁艺多, 等. 基于知识图谱的航天器数字资源建设[J]. 科技和产业, 2024, 24(2): 42-49.
|
| [5] |
|
| [6] |
|
| [7] |
王嘉宁. 面向预训练语言模型的提示调优方法[D]. 上海: 华东师范大学, 2024.
|
| [8] |
黄建勇. 基于深度学习的随机梯度下降优化算法研究[D]. 淮南: 安徽理工大学, 2024.
|
| [9] |
谢思静, 文鼎柱. 基于联邦分割学习与低秩适应的RoBERTa预训练模型微调方法[J]. 数据采集与处理, 2024, 39(3): 577-587.
|
| [10] |
黄云扬. 基于深度学习的知识图谱推理方法研究[D]. 成都: 电子科技大学, 2021.
|
| [11] |
|
| [12] |
彭亚飞, 杨凡德. 基于贝叶斯网络的航天任务态势不确定性融合[J]. 兵工自动化, 2022, 41(8): 41-46, 51.
|
/
| 〈 |
|
〉 |