SOM在广域网故障诊断中的应用  

Application of SOM in fault diagnosis of WAN

作　　者：陈彦德[1] 潘志松[2] 缪志敏[2] 周志杰[1] 孙永红[3] 

机构地区：[1]解放军理工大学通信工程学院 [2]解放军理工大学指挥自动化学院 [3]解放军73672部队

出　　处：《解放军理工大学学报（自然科学版）》2009年第1期17-22,共6页Journal of PLA University of Science and Technology(Natural Science Edition)

基　　金：国家自然科学基金资助项目(60603029);江苏省自然科学基金资助项目(BK2005009)

摘　　要：为了提高网络管理重要组成部分的故障管理能力,通过对性能数据和故障征兆的推理来确定故障源,针对WAN(w ide area netw ork)的故障传播特征提出了一种基于自组织特征映射SOM(self-organ izingm aps)的网络故障诊断模型,该模型包括SOM训练模块和故障实时诊断模块。仿真实验表明,采用该模型进行4种故障诊断,故障识别率达到95.82%,而误警率只有3.96%,说明在WAN中采用基于SOM的故障识别方法效果良好。In order to improve the ability of fault management,fault diagnosis was conducted on based on performance data and fault symptoms.A model of fault diagnosis based on self-organizing maps was proposed by analyzing the alarm propagation characters in the wide area network.The model was composed of two parts: self-organizing maps training by historical performance data and online fault diagnosis.The two simulation experimental results prove that the model achieves 95.82% of detection rate and 3.96% of false po...

关 键 词：广域网 故障管理 故障诊断 自组织特征映射 

分 类 号：TP393.2[自动化与计算机技术—计算机应用技术]