Abstract: To facilitate the transition of spacecraft testing mode from physical-experiments-based testing to physical-virtual-experiments-based testing,and to address the testing problems of long process cycle,high cost,great impacts on longevity and the difficulty of sharing of huge amount of data caused by the inconsistency between ground and in orbit,a Digital Twin (DT) and Multi-Agent Systems (MAS) based intelligent test architecture design method was proposed.An intelligent paradigm including physical space,data space and mathematical space for analyzing testing tasks was put forward,and an intelligent function modeling framework was built.The structures of three types of agents-test resource agent,test process agent and test service agent were designed,and the test flow optimization method for DT and MAS based spacecraft closed-loop life-cycle development was proposed.An intelligent test control and management software was developed to verify the feasibility and effectiveness of the intelligent test method by a spacecraft system-level test.By development and application of the intelligent test control and management platform,the time spent on product test design and implementation was reduced by 20%,and the precision of product test and tooling design was increased by 15%.

Key words: digital twin, multi-agent systems, intelligent test, architecture, spacecraft

摘要： 为推动航天器以物理试验为主导的模式向以物理试验与虚拟试验深度融合的模式转型,针对航天器试验过程中存在的周期长、成本高与其他制造环节海量数据共享困难等问题,提出基于数字孪生和多智能体的智能试验体系架构设计方法。该方法首先建立包含物理空间、数字空间和数学空间的智能试验任务分析范式,基于数字孪生和多智能体的智能试验体系功能建模框架;然后针对智能试验全要素、全过程、全业务的应用背景,设计试验资源、试验过程和试验服务3类智能体的结构模型,并对试验运行机制和技术流程进行优化。以某航天器系统级正样热试验为例搭建了智能试验管控平台,对该方法的可行性和有效性进行验证。结果表明,智能试验管控平台的应用,使得产品试验设计与实施总周期缩短20%、产品试验与工装设计精度提高15%。

关键词: 数字孪生, 多智能体, 智能试验, 体系架构, 航天器