卫星成像任务指令序列自适应生成方法研究
Research on Adaptive Generation of Instruction Sequence for Satellite Imaging Missions
王路桥 1李青山 1吴磊 1王璐 1杨晓龙 2李晓锋2
作者信息
- 1. 西安电子科技大学计算机科学与技术学院,西安 710071
- 2. 北京控制工程研究所,北京 100094
- 折叠
摘要
针对成像任务到指令序列转译时存在的任务失效风险,提出一种知识驱动的指令序列自适应生成方法.在指令构造阶段,将数据与逻辑相互解耦,并设计3种支持灵活修改的基本知识结构.在序列生成阶段,进一步引入图结构知识,并设计上下游压缩的序列生成及优化策略.该方法将关键数据要素从指令处理过程中分离,不仅降低了系统更新所引入的潜在风险,且提高了指令序列生成系统对需求变更的适应能力.采用历史成像任务序列进行仿真实验,结果表明,相比于现役指令处理方法,所提方法具有更高的准确率,并能有效优化指令序列长度.
Abstract
To address the risk of mission failure in imaging missions to instruction sequence translation,a method of knowledge-driven adaptive instruction sequence generation is proposed.During the instruction construction phase,data and logic are decoupled,and three flexible basic knowledge structures are designed.In the sequence generation phase,graph-based knowledge is introduced,and an upstream and downstream compressed sequence generation and optimization strategy is developed.This method separates critical data elements from the instruction translation process,thereby reducing potential risks from system updates and enhancing the adaptability to changing requirements.Historical imaging mission sequences are adopted to conduct simulation experiments.The results demonstrate that our method achieves higher accuracy and effectively optimizes the length of instruction sequence compared to existing methods.
关键词
指令上注/成像任务/指令构造/序列生成/自适应处理Key words
Instruction release/Imaging mission/Instruction construction/Sequence generation/Adaptive processing引用本文复制引用
基金项目
国家自然科学基金(U21B2015)
国家自然科学基金(61972300)
国家自然科学基金(62372351)
陕西省科协青年人才托举计划项目(20220113)
出版年
2024
NETL
NSTL
万方数据