«上一篇/Previous Article|本期目录/Table of Contents|下一篇/Next Article»

《计算机技术与发展》[ISSN:1006-6977/CN:61-1281/TN]

卷:

30

期数:

2020年09期

页码:

1-6

栏目:

智能、算法、系统工程

出版日期:

2020-09-10

文章信息/Info

Title:

A Strategy of Forecasting Station Traffic Based on Parallel Combination Model

文章编号:

1673-629X(2020)09-0001-06

作者:

冯培坤¹; 刘 杰²; 伍卫国³; 柴玉香¹; 张祥俊³

1. 西安交通大学 软件学院,陕西 西安 710000；2. 上海超级计算中心,上海 201203；3. 西安交通大学 计算机学院,陕西 西安 710000

Author(s):

FENG Pei-kun¹; LIU Jie²; WU Wei-guo³; CHAI Yu-xiang¹; ZHANG Xiang-jun³

1. School of Software,Xi’an Jiaotong University,Xi’an 710000,China; 2. Shanghai Supercomputing Center,Shanghai 201203,China; 3. School of Electronics and Information Engineering,Xi’an Jiaotong University,Xi’an 710000,China

关键词:

卡尔曼滤波; 支持向量机; 时序预测模型; 并联组合模型; 流量预测模型

Keywords:

Kalman filtering; support vector machine; time series prediction model; parallel combination model; traffic prediction model

分类号:

TP301. 6

DOI:

10. 3969 / j. issn. 1673-629X. 2020. 09. 001

摘要:

随着流量数据的增加,网络流量呈现出复杂多变的特点,为了方便站点的运营和网络的管理,需要对网络流量进行预测。 当前,网络流量预测多采用回归预测模型、卡尔曼滤波模型、神经网络模型、支持向量机模型等方法。 文中考虑高性能计算环境下作业管理系统日志流量的特点和预测需求等因素, 通过分析卡尔曼滤波算法和支持向量机的原理与缺点,提出了一种基于时间序列,结合卡尔曼滤波和支持向量机的并联组合模型,并对其进行了测试与分析。 结果表明,在相同的环境下,基于卡尔曼滤波算法和支持向量机的并联组合模型相比于卡尔曼滤波算法和支持向量机单个模型对流量的预测与实际流量值误差更小,预测结果也是可靠有效的,更适用于预测站点流量。

Abstract:

With the increase of traffic data,network traffic is characterized by complexity and diversity. In order to facilitate the operation of the site and network management,it is necessary to predict the network traffic. At present,the network traffic prediction mostly adopts regression prediction model, Kalman filter model, neural network model and support vector machine model. Considering the characteristics and forecasting requirements of job management system log traffic in high performance computing environment, by analyzing the principles and disadvantages of Kalman filtering algorithm and support vector machine,we propose a parallel combination model based on time series combined with Kalman filter and support vector machine,which is tested and analyzed. The test results show that in the same environment,the parallel combination model based on Kalman filtering algorithm and support vector machine has less error in predicting the flow and actual flow value compared with the single model based on Kalman filtering algorithm and support vector machine,and the prediction results are also reliable and effective,which is more suitable for predicting the site flow.

常用功能

更新日期/Last Update: 2020-09-10