摘要
当前,通过构建故障树进行故障定位的实时性较差、难以适应紧急故障,且构建和分析故障树需要的专业知识难以固化、历史发生的故障案例并未被有效利用。针对上述问题,文章提出了一种基于故障树分析的应用大语言模型的故障定位改进方法。首先,基于历史上发生的故障案例或预先分析的故障知识构建故障知识库,实现故障相关专业知识的固化,包括通过故障树分析得到的故障原因及对应的故障现象;然后,利用大语言模型的检索、深度理解、语言归纳能力,结合故障知识库根据实际的故障现象分析出可能的故障原因,从而准确快速地实现故障定位。根据上述原理搭建了故障定位系统,并进行了试验验证,其故障定位的准确率达到85%,验证了所提方法的可行性。
Addressing the poor real-time performance in fault localization based on constructing the fault tree and the challenge of solidifying the professional knowledge required for constructing the fault trees,an improved fault localization method based on fault tree analysis using large language models is proposed in this paper.Firstly,a fault knowledge base is constructed from historical fault cases or pre-analyzed fault knowledge to solidify the relevant professional knowledge,including the fault causes and corresponding fault phenomena obtained through fault tree analysis.Then,utilizing the large language models’capabilities of retrieval,deep comprehension and language generalization,the possible fault causes can be analyzed based on actual fault phenomena and the knowledge base.A fault localization system is constructed based on the above principle and experiments are conducted.The results show that the accuracy of fault localization reaches 85%,validating the feasibility of the proposed method.
作者
肖雪迪
何宇
张伟
周寻
安洲
XIAO Xuedi;HE Yu;ZHANG Wei;ZHOU Xun;AN Zhou(Beijing Institute of Spacecraft System Engineering,Beijing 100094,China)
出处
《航天器工程》
CSCD
北大核心
2024年第5期15-21,共7页
Spacecraft Engineering
关键词
航天器故障定位
大语言模型
知识库
spacecraft fault localization
large language model
knowledge base
