摘要： 传统BP神经网络对网络流量时间序列预测精度低和泛化能力弱。为此,提出一种新的优化BP神经网络的方法。通过小波包分解对网络流量进行多频段序列分解,并采用飞蛾纵横交叉混沌捕焰算法优化的神经网络,对各分解后的子序列进行预测,叠加各子序列的预测值,重构获取实际预测结果。仿真结果表明,与传统BP神经网络预测方法相比,该方法能捕获网络流量的变化规律,具有较好的预测精度、稳定性和泛化能力。

关键词: 飞蛾捕焰算法, 网络流量预测, 小波包分解, 神经网络, 预测计算

Abstract: Traditional BP neural network has low prediction accuracy and weak generalization ability for network traffic time series.Therefor,a new method for optimizing BP neural network is proposed.The network traffic is decomposed into multi-channel series through wavelet packet decomposition,and the neural network optimized by moth crisscross chaos flame capturing algorithm is employed to forecast the decomposed sub-series.The predicted values of each sub-series are superimposed and the actual prediction results are obtained by reconstruction.Simulation results show that compared with the traditional BP neural network prediction method,this method can capture the variation law of network traffic,and it has good prediction accuracy,stability and generalization ability.

Key words: moth-flame algorithm, network traffic prediction, wavelet packet decomposition, neural network, prediction calculation

中图分类号: 

• TP393