基于知识图谱的装备故障诊断技术  

Equipment fault diagnosis technology based on knowledge graph

作　　者：赵永亮 于倩 邓博 韩丽君 ZHAO Yongliang;YU Qian;DENG Bo;HAN Lijun(Xi’an Jiutian Digital Intelligence Information Technology Co.,Ltd.,Xi’an 710086,China)

机构地区：[1]西安九天数智信息科技有限公司,陕西西安710086

出　　处：《电子设计工程》2022年第9期125-129,共5页Electronic Design Engineering

基　　金：陕西省重大科技创新专项资金项目(2016ZKC02-04);陕西省技术创新引导专项(基金)(2020CGXNX-039)。

摘　　要：传统的故障诊断方法为建立解析数学模型,该模型只能在装备发生故障后才能进行诊断分析,无法满足现在武器装备的故障检测需求。针对这一问题,文中以多层次知识图谱模型为基础,使用贝叶斯网络进行故障检测和诊断并完成建模,模型可以实现武器装备的状态变化检测及故障判断。实验结果表明,该模型可以通过状态的改变进而对故障类型进行判断。在对比实验中,文中所提算法模型的准确率与其他对比算法相比提高了2.5%、4.2%和5.6%,这说明该算法可以对装备进行故障诊断且综合性能良好。Traditional fault diagnosis methods establish analytical mathematical model,the model can only be diagnosed and analyzed after the equipment failure,which can not meet the fault detection requirements of weapon equipment. In order to solve this problem,based on the multi-level knowledge map model,Bayesian network is used to detect and diagnose the fault and complete the modeling. The model can realize the state change detection and fault judgment of weapon equipment. The experimental results show that the model can judge the fault type by changing the state. In the contrast experiment,the accuracy of the proposed algorithm model is improved by 2.5%,4.2% and 5.6% compared with the contrast algorithm,which shows that the algorithm can diagnose the equipment fault and has good comprehensive performance.

关 键 词：知识图谱 贝叶斯网络 故障检测 复杂装备 数据分析 状态检测 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]