贝叶斯网络支持下的计算机硬件系统故障诊断分析  

Fault diagnosis analysis of computer hardware system supported by Bayesian network

作　　者：余鸣 余哲源 YU Ming;YU Zheyuan(Information Technology Department of Qujing Vovationali and Technical College,Qujing,Yunnan 655000,China;School of Materials Science and Engineering,Shenyang Jianzhu University,Shenyang 110168,China)

机构地区：[1]曲靖职业技术学院信息技术系,云南曲靖655000 [2]沈阳建筑大学材料科学与工程学院,沈阳110168

出　　处：《计算机应用文摘》2024年第14期152-154,共3页Chinese Journal of Computer Application

摘　　要：文章首先分析了硬件故障的特性,提出了将概率图模型作为解决方案的思路,接着详细介绍了如何构建反映系统拓扑和组件间依赖关系的贝叶斯网络模型,最终实施了包括数据采集、学习和预测在内的完整流程。服务器集群实际验证表明,该模型能够以高达89%的准确率进行故障诊断,且预测时间最多可提前1.5 h,显示出明显优于传统规则方法的智能化和动态化特点。This article first analyzes the characteristics of hardware failures and proposes the idea of using probability graph models as a solution.Then,it elaborates on how to build a Bayesian network model that reflects the system topology and dependency relationships between components.Finally,a complete process including data collection,learning,and prediction is implemented.The actual validation of server clusters shows that the model can diagnose faults with an accuracy of up to 89%,and the prediction time can be advanced by up to 1.5 hours,demonstrating significantly better intelligent and dynamic characteristics than traditional rule methods.

关 键 词：计算机硬件 系统故障 智能诊断 贝叶斯网络 概率图模型 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]