基于递归神经网络的网络流量组合预测模型

为了提高网络流量的预测精度,提出了一种基于Elman递归神经网络,小波和自回归的网络流量组合预测模型.对流量时间序列进行小波分解,得到小波变换尺度系数序列和小波系数序列,对具有平稳特征的尺度系数序列用AR模型进行预测:而对体现了网络流量非线性、非平稳特性的小波系数序列使用Elman递归神经网络进行预测,最后通过Mallat算法重构得到网络流量的预测值.

Combination prediction model of network traffic based on recurrent neural networks

For improving the prediction accuracy of network traffic,a new combination prediction model is proposed based on the in-tegration of recurrent neural networks,wavelet and the auto-regression(AR).First,the traffic time series are decomposed to the scaling coefficient series and wavelet coefficient series.The scaling coefficient series is predicted by AR model,while the wavelet coefficient series is predicted by Elman recurrent neural networks.The predicted results of the two series are restructured by Mallat arithmetic at last.