摘要
为解决卫星故障诊断系统研制周期长、重复性工作量大的缺点,针对基于平行系统和数字卫星概念设计的卫星故障诊断系统,提出一种智能研制技术。建立包含参数、接口、原理模型、误差模型、故障模型和三维模型的数字部件库,利用数字装配技术实现对卫星的规范化描述,基于逻辑维分形结构将源代码书写规范分为确定性规则、有限规则和推理规则,采用机器学习的方法完成源代码书写规则的推理和实施,实现包括子系统部件、姿轨控软件、星务软件、耦合场计算、其他文件和地面测控系统等的数字卫星、遥测遥控系统与配套工具的源代码自动书写、编译及部署。针对多颗卫星进行故障诊断系统研制,系统运行结果表明了智能研制技术在提高系统研制效率方面的可行性和有效性。
To deal with the problem of long development cycle and heavily repetitive program workload of traditional satellite fault diagnosis system,an intelligent development technology was proposed to develop satellite fault diagnosis system based on parallel system and digital satellite.A digital device database was established including device parameters,interfaces,principle models,error models,fault models and 3-dimension models.The standard description of the digital satellite was determined by digital assembly.Based on the fractal structure of logic dimension,the source code generation rules of the satellite fault diagnosis system can be divided into three types that are deterministic rules,limited rules and rules worked out by inference.Machine learning is applied for reasoning and establishing source code generation rules,and these rules are used to accomplish source code generation,compilation and disposition of parts of sub system,attitude and orbit control software,satellite operation software,coupling field computation,other documents and ground test and control system.Several satellite fault diagnosis systems are developed according to different satellites,whose performances and effectiveness in improving system development efficiency are proved by simulation results.
出处
《上海航天》
CSCD
2017年第3期52-59,共8页
Aerospace Shanghai
基金
国家重点基础研究发展计划(973计划)资助(613237XX)
关键词
卫星
故障诊断
平行系统
智能研制
数字装配
机器学习
逻辑维
分形
satellites
fault diagnosis
parallel system
intelligent development
digital assembly
machine learning
logical dimension
fractal
