文献类型：期刊文献

中文题名：基于最小熵翻卷积的网络故障特征提取仿真

英文题名：Simulation on Network Fault Feature Extraction Based on Minimum Entropy Turn Convolution

作者：马涛[1];秦轶翚[1];魏绍谦[1]

第一作者：马涛

机构：[1]北京联合大学师范学院

第一机构：北京联合大学师范学院

年份：2015

卷号：32

期号：4

起止页码：269-272

中文期刊名：计算机仿真

外文期刊名：Computer Simulation

收录：CSTPCD;;北大核心:【北大核心2014】；CSCD:【CSCD_E2015_2016】；

语种：中文

中文关键词：逆滤波器;信噪比;最小熵翻卷积;故障特征

外文关键词：Inverse filter; Signal to noise ratio ; Minimum entropy turn convolution; The fault feature

摘要：传统的基于FRFT网络故障特征提取方法当网络信号发生突变时,由于受到噪声和信号衰弱的影响,导致网络故障特征极其微弱,并且网络的拓扑结构和权值分布成非线性映射,将信号简单排列成矩阵,无法有效实现对网络故障特征的提取。提出一种基于小波滤波以及最小熵翻卷积的网络故障特征提取方法,将突变信号在与之相邻尺度上的小波系数直接相乘,依据阈值对噪声中的网络故障信息进行采集并过滤噪声,使获取的小波系数信噪比大大增强。将突变信号小波变换值在几个尺度上进行计算,实现网络故障特征的初提取。获取一个逆滤波器,通过网络输出恢复网络输入信号,依据解卷积后获取的序列对可能估计值的最优解进行计算,求出逆滤波器矩阵,分析了最小熵归迭代算法的具体实现过程。仿真结果表明,所提方法具有很高的准确性。
A network fault feature extraction method is presented based on the wavelet filtering and the minimum entropy convolution, the mutation signal is directly multiplied by the adjacent scale wavelet coefficient, according to the threshold of network fault information acquisition and filtering of noise, the wavelet coefficients are obtained to greatly enhance the signal - to - noise ratio. The wavelet transform values of the mutation signals are calculated of in geometric scales, and netwoTk early failure feature extraction is realized. An inverse filter is designed, the network input signal is restored by output signals, the optimal solution is calculated for possible estimated value according to the sequence of convolution, the inverse filter matrix is obtained, and the minimum entropy for concrete implementa- tion process of iterative algorithm is analyzed. The simulation results show that the proposed method has high accuracy.