基于主成分分析的网络时延特征数据提取仿真

Simulation of Network Delay Feature Extraction Based on Principal Component Analysis

为了能够快速提供网络时延特征数据,需要对其进行特征数据提取。但当前特征数据提取过程中,普遍存在着特征数据提取所需完成时间过长、提取误差率较大、成本消耗较高等问题。提出基于主成分分析法的网络时延特征数据提取方法。通过对网络时延特征数据进行分析,采用最小-最大标准化方法对特征数据进行模糊化处理,获取特征数据的模糊规则匹配度,将模糊规则匹配度在各个分类规则上进行归一化处理得到特征数据分类匹配度,求出特征数据规则权重值,利用特征数据样本分类健全度对网络时延特征数据规则权重值进行分类。利用主成分分析法获取特征数据特征向量,采用Relief算法对特征数据向量进行提取,实验结果表明,所提出方法特征数据提取所需完成时间短、提取误差率较小、成本消耗较低。

In order to provide the network delay feature data quickly,it is necessary to extract the feature data.Traditional feature data extraction methods have long extraction time,large the error rate and high cost consumption.Therefore,a network delay feature data extraction method based on principal component analysis was put forward.Based on the analysis of network delay feature data,the minimum-maximum standardization was used to blur the feature data,so that the fuzzy rule matching degree of feature data was obtained.The matching degree of fuzzy rule was normalized on each classification rule to get the classification matching degree of feature data,so as to find weight value of feature data rule.The integrity degree of feature data sample classification was used to classify the weight values of network delay feature data rule.Moreover,the principal component analysis was used to obtain the feature vector of feature data.Finally,feature data vector was extracted by Relief algorithm.Simulation results show that the proposed method needs short time to extract feature data.Meanwhile,the error rate and cost are low.