摘要
网络流量是一种典型的时间序列数据,具有很强的滞后性和后效性。针对当前滞后阶数确定方法存在局部最优,耗时长等缺陷,提出一种网络流量组合预测方法(GS-GA-LSSVM)。首先采用地统计学(GS)快速确定网络流量的最优滞后阶数,然后根据滞后阶数对网络流量进行重构,最后采用遗传算法(GA)优化最小二乘支持向量机LSSVM(least square support vector machine)对网络流量进行建模预测。仿真结果表明,GS-GA-LSSVM对网络流量的预测精度优于参比模型,更能反映网络流量复杂的动态变化规律。
Network traffic is a typical time series data which has very strong posteriority and after effectiveness.This paper puts forward a combinatorial network traffic forecasting method（GS-GA-LSSVM） to solve the defects that current lag order determination method is of local optimal and time-consuming.First,the geostatistics is used to fast determine the optimal lag order of network traffic,then the data will be reconstructed according to the determined order,finally the least squares support vector machine（LSSVM） optimised by the genetic algorithm（GA） is used for modelling the forecast.Simulation results show that the forecasting accuracy of the proposed method on network traffic is better than the reference models,and reflects the complex dynamic change rule of network traffic better.
出处
《计算机应用与软件》
CSCD
北大核心
2012年第9期139-142,共4页
Computer Applications and Software
基金
山东省自然科学基金项目(ZR2011FL006)
