基于时序图模式匹配的航天器故障诊断算法  

A Spacecraft Fault Diagnosis Algorithm Based on Temporal Graph Pattern Matching

作　　者：罗瑞欣 刘显敏[1] 高宇鹏 梁寒玉 张妍 LUO Ruixin;LIU Xianmin;GAO Yupeng;LIANG Hanyu;ZHANG Yan(Faculty of Computing,Harbin Institute of Technology,Harbin 150001,China;Beijing Institute of Control Engineering,Beijing 100094,China)

机构地区：[1]哈尔滨工业大学计算学部,哈尔滨150001 [2]北京控制工程研究所,北京100094

出　　处：《宇航学报》2025年第2期262-271,共10页Journal of Astronautics

基　　金：国家自然科学基金项目(U1811461,62022013,12150007,62103450,61832003,62272137);科技部重点研发项目(2021YFB1715000);黑龙江省高校大学专项科研资金项目(2022-KYYWF-1122)。

摘　　要：故障诊断是支撑航天器在轨健康运行的有效手段,其主要任务是检测故障并判断其发生的具体位置和原因。由于结构复杂、工作环境极端等原因,航天器在轨监测数据与实际故障间关系难以获取,仅通过数据分析来实现故障诊断的方案可行性很低。对于同时利用知识图谱和监测数据共同完成故障诊断的思路,现有研究工作大都未考虑故障与一段时间内数据异常模式间的关系。为解决上述问题,提出了基于时序图模式匹配的航天器故障诊断算法,利用时序图表示连续时间片段内数据异常间的时序关系,支持描述更精细且更稳定的异常模式,使新算法实现更精准且更高效的航天器故障诊断。在航天数据上的实验结果表明,新算法具有更高的故障诊断精度,且时间代价较低。Fault diagnosis serves as an effective method for ensuring the healthy operation of spacecraft in orbit.Its primary objective is to identify faults and pinpoint their specific locations and causes.Given the intricate structure and extreme operational environment of spacecraft,obtaining the relationship between on-orbit monitoring data and actual faults poses a challenge.Consequently,relying solely on data analysis for fault diagnosis proves to be highly infeasible.Regarding the approach of utilizing knowledge graphs and monitoring data for fault diagnosis,much of the existing research overlooks the temporal relationship between faults and data anomaly patterns.To address these issues,an algorithm for spacecraft fault diagnosis based on time series graph pattern matching is introduced.This algorithm employs time series graphs to depict the temporal relationships among data anomalies across continuous time segments,enabling the description of more refined and stable anomaly patterns.As a result,the new algorithm facilitates more accurate and efficient fault diagnosis of spacecraft.Experimental results utilizing aerospace data demonstrate that the new algorithm exhibits higher accuracy in fault diagnosis and reduced time costs.

关 键 词：异常模式 故障诊断 时序关系 航天器 

分 类 号：V241.9[航空宇航科学与技术—飞行器设计]