基于卷积神经网络的网络故障诊断模型  

A Network Fault Diagnosis Model Based on Convolution Neural Network

作　　者：李酉戌 

机构地区：[1]上海理工大学光电信息与计算机工程学院,上海200093

出　　处：《软件导刊》2017年第12期40-43,共4页Software Guide

摘　　要：对于网络运行过程中产生的海量日志信息,传统故障诊断方法很难进行实时而全面的日志分析。针对该问题,提出一种基于卷积神经网络(Convolutional Neural Network,CNN)的网络故障诊断模型,利用Skip-gram模型进行词向量训练,并将词向量作为卷积神经网络输入,最终通过Softmax回归进行分类。实验结果表明,该模型可以有效处理网络故障诊断任务,且优于传统机器学习方法,对于网络运行日志的故障诊断准确率可达73.2%以上。In order to solve the problem,this paper presents a network fault diagnosis model based on Convolutional Neural Network（CNN）,which is difficult to carry out real-time and comprehensive log analysis for the mass log information generated by the network running process.Diagnostic model,the skip-gram model for word vector training,and the word vector as a neural network input,and ultimately through the Softmax regression classification.The experimental results show that the model can effectively deal with the task of network fault diagnosis and is superior to the traditional machine learning method.The fault diagnosis accuracy of the network running log can reach more than 73.2%.

关 键 词：日志信息 故障诊断 卷积神经网络 词向量 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]