基于BPSO-RBF神经网络的网络流量预测

NETWORK TRAFFIC PREDICTION MODEL BASED ON BPSO-RBFNN

为了提高网络流量的预测精度,针对网络的时变性和混沌性,提出一种反向学习粒子群优化神经网络的网络流量预测模型(BPSO-RBFNN)。首先将网络流量样本输入到RBF神经网络进行学习,采用引入反向学习机制的粒子群算法优化参数,然后建立网络流量预测模型,最后采用仿真实验对模型性能进行分析。结果表明,BPSO-RBFNN可以描述网络流量的时变性、混沌性变化趋势,网络流量预测精度得以提高,具有较好的实际应用价值。

In order to improve the prediction accuracy of network traffic,in this paper we propose a network traffic prediction model（BPSO-RBFNN）,which is based on neural network optimised by the opposition-based learning particle swarm optimisation.First,we inputthe network traffic sample to RBF neural network for learning,and introduce particle swarm optimisation of opposition-based learningmechanism to optimise the parameters,then we build network traffic prediction model,and finally use simulation experiment to analysemodel’s performance.Results show that the BPSO-RBFNN can describe the variation trend of time-varying property and chaotic property ofthe network traffic,and the prediction accuracy of network traffic can be improved,it has higher practical application value.