基于混沌粒子群优化SVR的网络流量预测  

Network Traffic Prediction Based on SVR Optimized by Chaos Particle Swarm Optimization Algorithm

作　　者：王治[1] 

机构地区：[1]九江学院信息科学与技术学院,江西九江332005

出　　处：《计算机仿真》2011年第5期151-154,共4页Computer Simulation

基　　金：省教学改革课题项目资助(JXJG-O7-17-50)

摘　　要：计算机网络流量预测对于网络流量的控制和调整,进而提高网络性能和服务质量起着重要的作用。当前传统计算机网络流量方法预测精度较低,仅为85%左右。针对网络流量的非线性和时变特性,难以准确实现网络流量评估,为了解决此问题,提出基于混沌粒子群优化SVR的网络流量预测方法。支持向量回归算法(SVR)是一种用于趋势预测的支持向量机模型,能找到全局最优解。然而,SVR参数的选择对回归模型优化起着决定性作用。采用混沌粒子群优化算法(CPSO)优化支持向量参数,通过建立混沌粒子群优化SVR的网络流量预测模型。仿真结果表明,混沌粒子群优化SVR网络流量预测模型能力强、效果好。Prediction of computer network traffic is very important to control and adjust network traffic,and improve the performance of network and quality of service.At present,prediction accuracy of traditional computer network traffic method is low,only about 85%.Thus,the network traffic evaluation is difficult to realize.In order to solve the problem,network traffic prediction based on SVR optimized by chaos particle swarm optimization algorithm is presented.Support vector regression algorithm is a kind of support vector machine model used in trend prediction,which can find globally optimal solution.However,the choice of SVR parameters has a decisive effect on generalization ability of the regression model.There exists computational instability in genetic algorithm,and particle swarm optimization algorithm falls into local extremum easily in spite of the simple of the algorithm.In the paper,SVR parameters optimized by chaos particle swarm optimization algorithm is presented,and network traffic prediction model based on SVR optimized by chaos particle swarm optimization algorithm is created.SVR optimized by chaos particle swarm optimization algorithm has higher network traffic prediction ability than SVR optimized by genetic algorithm and SVR optimized by particle swarm optimization algorithm.

关 键 词：支持向量回归算法 混沌粒子群优化 网络流量 预测 

分 类 号：TP393.07[自动化与计算机技术—计算机应用技术]