基于改进的HHT的网络流量预测算法

Network traffic prediction algorithm based on improved HHT

由于传统的网络流量预测算法对网络流量的线性程度要求高,而现实网络流量多是非线性的,提出一种基于改进的HHT和SVR网络流量预测算法。针对HHT方法存在的拟合过冲和端点飞翼的缺陷,提出一种求取均值曲线的方法对HHT进行改进。实验结果表明,利用该方法得到的包络线效果更好,均值点更加准确,基本消除了过冲的影响,对端点效应有明显改善。

As the traditional linear network traffic prediction algorithms for network traffic require high linearity ,but the reality network traffic is nonlinear ,an algorithm for network traffic prediction which based on improved HHT and SVR was proposed . HHT method has the problems of fitting overshoot and endpoint wing ,for this ,a method for obtaining the mean curve to im‐prove HHT was proposed .Experimental results show that the effects of the envelopes obtained by the proposed method are bet‐ter and the average points are more accurate .The proposed method fundamentally eliminates the overshoot influence ,and the end effect is significantly improved .