基于自适应学习算法的智能代理网络故障监测

Network Fault Detecting by the Intelligent Agent Based on Adaptive Learning Algorithm

提出一种基于自适应学习算法的故障监测智能代理，通过分段测量抽取描述网络正常行为的ＭＩＢ变量值并检测偏差，对学习获得的信息经由贝叶斯图加以组合，从而鉴别未知的或不可预见的故障．实验结果表明，智能监测代理系统能够在故障发生以前检测网络异常行为．

An intelligent agent for fault detecting is proposed based on adaptive learning algorithm. By segmentaly measuring, MIB variable for describing network the normal behavior is extracted and the deviation is detected. This information is combined by the means of Bayesian Graph so as to identify unknown or unpredictable faults. The experimental results show that the intelligent agent can detect network abnormal behavior before a fault actually occurs.