基于双向长短期记忆循环神经网络的网络流量预测

NETWORK TRAFFIC PREDICTION BASED ON BILSTM RECURRENT NEURAL NETWORK

针对长短期记忆循环神经网络在对时间序列进行学习时存在早期特征记忆效果差、难以充分挖掘整个网络流量特征等问题,提出一种基于双向长短期记忆循环神经网络的网络流量预测方法,以提高网络流量预测的准确性。对网络流量序列进行双向学习,避免单向学习导致较早学习部分特征提取和记忆效果差的问题。同时双向学习可以充分挖掘网络流量天与天之间双向的特征,完整地学习到网络流量的整体特征。仿真实验结果表明,改进后的方法相比原方法具有更好的预测效果。

For the LSTM recurrent neural network,when learning the time series,the early feature memory effect is poor,and it is difficult to fully exploit the characteristics of the entire network traffic.This paper proposes a network traffic prediction method based on BiLSTM recurrent neural network to improve the accuracy of network traffic prediction.The method performed two-way learning on the network traffic sequence,avoiding the problem that the one-way learning leads to poor feature extraction and memory effect in the early learning.Two-way learning could fully exploit the two-way characteristics of network traffic between day and day,and fully learn the overall characteristics of network traffic.The simulation results show that the improved method has better prediction effect than the original method.