%0 Journal Article
%T 基于Morlet-SVR和ARIMA组合模型的网络流量预测
%A 赵建龙
%A 曲桦
%A 赵季红
%A 戴慧珺
%J 北京邮电大学学报
%D 2016
%R 10.13190/j.jbupt.2016.02.011
%X 摘要 针对网络流量的非线性和多维度动力学特性，结合小波多尺度分析的能力，提出了基于Morlet小波核函数的支持向量机回归算法（Morlet-SVR）和自回归积分滑动平均模型（ARIMA）的组合模型预测网络流量.采用Morlet-SVR和ARIMA分别预测通过Mallat小波分解和单支重构得到的近似信号和多尺度细节信号，最后通过线性叠加得到最终预测结果.通过仿真实验分别对比分析了基于径向基核函数的支持向量机回归算法和ARIMA预测模型，通过3种误差评估得知该组合模型具有更高的预测精度.</br>According to the nonlinear and multi-dimensional dynamic characteristics of network traffic, combined with the ability of multi-scale wavelet analysis, a comprehensive forecasting model based on Morlet-support vector regression (Morlet-SVR) and auto regressive integrated moving average (ARIMA) was proposed, in which Morlet-SVR and ARIMA are employed to forecast the approximate signal and the multi-scale detail signals respectively by use of Mallet wavelet decomposition and single reconstruction. The final prediction result is obtained by linear superposition of the layers. Simulations give out comparisons with radial basis function-support vector regression and ARIMA model respectively, the proposed model shows higher prediction accuracy by comparison with three error evaluation measurements.
%K 流量预测
%K 小波核函数
%K Morlet支持向量机回归算法
%K 自回归积分滑动平均模型&lt
%K /br&gt
%K Key words： traffic prediction wavelet kernel function Morlet-support vector regression auto regressive integrated moving average
%U http://journal.bupt.edu.cn/CN/abstract/abstract2878.shtml