摘要
为了更准确的预测话务量,提出了一种以粒子群优化算法为基础的,通过多样性度量指标控制种群特征的改进粒子群优化算法(MPSO),用于最小二乘支持向量回归机(LS-SVR)超参数的寻优,分析影响话务量的相关因素,选取合适的样本,利用优化后的LS-SVR模型对移动话务量进行预测。与标准LS-SVR预测算法和PSO优化后的LS-SVM算法进行比较,实验结果表明,本文的预测方法具有更好的收敛性和更高的预测精度。
The Modified Particle SwannOptirnization （MPSO） algorithm that can control the characteristics of population by diversity metrics is presented in this paper based on Particle Swarm Optimization （PSO） algorithm in order to predict the telephone traffic accurately. The hyper parameter of Least Squares Support Vector Regression （IS- SVR） is optimized via MPSO algorithm, analysis the various factors affecting telephone traffic and select suitable samples, and then the traffic of mobile company is forecasted by the optimized LS - SVR model. The predicted result is compared with the method of standard LSSVR and PSO - LSSVR prediction algori＇thm, dae experimental .res.ults shov~ that MPSO - LSSVR forecast method has better convergence and hgher prediction accuracy...
出处
《激光杂志》
CAS
CSCD
北大核心
2013年第2期35-36,共2页
Laser Journal
基金
中国移动通信集团新疆有限公司研究发展基金项目(项目编号:XJM2011-11)
关键词
粒子群优化算法
最小二乘支持向量回归机
话务量预测
超参数
预测精度
particle .swarm op, tirnition .algorithm
leadt squaressuport vector regression
t lephone traffic focastmg
hyper paramneteri pr.etiction accacy
