摘要
研究网络故障诊断,针对保证网络安全可靠问题。网络故障诊断是一个复杂的过程,网络故障属性数量多且属性之间存在着大量冗余信息,传统故障诊断方法不能有效消除冗余信息,从而使网络故障诊断的准确率低。为了提高网络故障诊断的准确率,提出了一种粗糙集(RS)BP神经网络(BPNN)的网络故障诊断方法(RS-BPNN)。利用RS理论对网络故障属性集进行属性约简,消除属性之间冗余信息,使属性独立。将约简后的决策属性作为BPNN输入,通过BPNN非线性自学习,提高网络故障诊断的准确。通过RS-BPNN对网络故障数据进行仿真,仿真表明,相比传统网络故障诊断方法,RS-BPNN加快了网络故障诊断的速度,提高了网络故障诊断准确性,证明是一种有效的网络故障诊断工具。
Network fault diagnosis problem is studied.Because network fault attributes have too many redundant,the traditional network fault diagnosis cannot delete the redundant attributes.To improve there fault diagnosis accurate,a fault diagnosis method is put forward based the BP neural network and rough set.Firstly,attribute sets are reduced by rough sets theory,then reduction attributes are used as the input variables of BP neural network,which accelerates the BP neural network training speed and improves the network fault diagnosis accuracy.The results show that the proposed method has fast convergence speed and high accuracy.
出处
《计算机仿真》
CSCD
北大核心
2011年第3期186-189,共4页
Computer Simulation
关键词
粗糙集
神经网络
故障诊断
Rough set（RS）
Neural network（NN）
Fault diagnosis
