一种改进的小波神经网络的网络流量预测算法  

A Network Traffic Prediction Algorithm Based on Wavelet Function and Neural Network

作　　者：陈振伟[1] 王茜[1] 黄继红[1] 

机构地区：[1]皖西学院计算机科学与技术系,安徽六安237012

出　　处：《皖西学院学报》2010年第2期36-38,共3页Journal of West Anhui University

摘　　要：为了有效地对网络进行维护,提高网络性能,预知网络流量可以提前对网络出现的问题采取应对策略,从而对用户提供更好的服务。在神经网络预测模型中把隐含层的传递函数用小波函数替换,并采用共轭梯度下降算法,建立了一个小波神经网络的网络流量预测模型。通过实际流量数据对模型进行仿真,结果显示该模型与神经网络预测模型相比,该网络具有良好的预测效果,网络训练时间短,有效地提高了训练速度。To maintain network availably,and improve the performance of network,the corresponding strategy could be adopted in advance by the network traffic prediction,thus better service may be offered to users.In this paper,with introducing the wavelet function into the neural network traffic prediction model,and conjugate gradient descent algorithm,a wavelet neural network model used to predict the network traffic is proposed in this paper.Simulations are performed on the model by use of some real network traffic data,and the simulation results demonstrate that the model has better prediction effect and much higher training speed,compared with the original Network Traffic Prediction model.

关 键 词：神经网络 小波函数 网络流量 共轭梯度 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]