摘要
为了提高网络流量的预测精度,针对ESN网络预测结果易受谱半径SR、神经元个数N、输入单元尺度IS、连接稀疏度SD以及嵌入维数p等参数的影响,利用IWO算法收敛速度快的特点对ESN模型参数进行优化。研究结果表明,IWO-ESN可以有效提高网络流量的预测精度,同时可以对ESN网络模型参数和嵌入维数p进行自适应选择,为网络管理优化和拥堵控制提供决策依据。
In order to improve the accuracy of network traffic prediction,the prediction results for ESN network are susceptible to parameters such as parameter spectrum radius SR,neuron number N,input unit scale IS,connection sparse SD,and embedded dimension p,etc..The parameters of ESN model are optimized by using the characteristics of fast convergence speed of IWO algorithm.The results show that IWO-ESN can effectively improve the prediction accuracy,realize the adaptive selection of ESN model parameters,and embed dimension p,which can provide decision-making basis for network management optimization and congestion control.
作者
白亚秀
BAI Ya-xiu(School of Electronics and Information Engineering,Ankang University,Ankang 725000,Shaanxi Province,China)
出处
《信息技术》
2019年第7期110-115,共6页
Information Technology
关键词
入侵杂草优化算法
回声状态网络
网络流量
谱半径
连接稀疏度
intrusive weed optimization algorithm
echo state network
network traffic
spectral radius
sparse degree
