一种网络流量预测的小波神经网络新模型

A New Wavelet and neural network model of network traffic forecast

文中提出了一种将小波变换与相关性计算与人工神经网络相结合进行网络流量预测的新模型。首先对流量时间序列进行小波变换得到不同尺度下的细节信号和近似信号。对近似信号运用相关性的计算确定其自相关程度。根据相关程度确定神经网络的输入与输出,构造神经网络并加以训练。对细节信号采取传统的ARMIA预测。最后用小波重构得到最终的流量预测值。模拟仿真表明,该模型具有较高的预测效果。

In the papar a new network traffic prediction model which combines the wavelet transform ,the correlation calculation and neural network is presented. First, the time series is used wavelet transform we can get detail signal and approximation signal at different scales. By Correlation calculation get the Approximation signal’s degree of correlation. According to the degree of correlation determine the input and output of the neural network, an artificial neural network is established and trained. Detail signal is used ARIMA forecast. Finally, wavelet reconstruction is used to get the final the traffic value of predictable. The simulation results show that this model is more successful than traditional methods in network traffic prediction.