农业云视频平台虚拟机负荷预测半监督偏最小二乘法模型

为了优化基础设施资源的效率,农业云视频平台虚拟机布局算法需要了解虚拟机当前和未来的资源工作效率,尽可能准确地预知下一步工作,如服务部署,虚拟机的部署、迁移或停止.然而,通常在预测中使用的样本非常小,可用于分析的数据有限.因此,该文研究设计了一个考虑时间因素,基于小数据集学习的滑动窗口模型.此外,鉴于现有的预测算法仍然有很大的改进误差率的空间,该文中采用基于滑动窗口与最小二乘法和半监督学习的数学方法相结合,提出了一种半监督偏最小二乘法(semi-supervised partial least squares,SS-PLS)的方法来计算上述预测.该文中,分析了在虚拟机使用SS-PLS负荷预测的可...

Full description

Saved in:
Bibliographic Details
Published in农业工程学报 Vol. 33; no. z1; pp. 225 - 230
Main Author 高万林 胡慧 徐东波 张港红
Format Journal Article
LanguageChinese
Published 中国农业大学信息与电气工程学院,北京,100083 2017
Subjects
负荷;预测;算法;半监督学习;虚拟机;偏最小二乘法;滑窗
algorithms
forecasting
负荷
半监督学习
sliding-windows
滑窗
算法
虚拟机
virtual machine
偏最小二乘法
load
预测
semi-supervised learning
partial least squares
Online AccessGet full text
ISSN1002-6819
DOI10.11975/j.issn.1002-6819.2017.z1.034

Cover

More Information
Summary:为了优化基础设施资源的效率,农业云视频平台虚拟机布局算法需要了解虚拟机当前和未来的资源工作效率,尽可能准确地预知下一步工作,如服务部署,虚拟机的部署、迁移或停止.然而,通常在预测中使用的样本非常小,可用于分析的数据有限.因此,该文研究设计了一个考虑时间因素,基于小数据集学习的滑动窗口模型.此外,鉴于现有的预测算法仍然有很大的改进误差率的空间,该文中采用基于滑动窗口与最小二乘法和半监督学习的数学方法相结合,提出了一种半监督偏最小二乘法(semi-supervised partial least squares,SS-PLS)的方法来计算上述预测.该文中,分析了在虚拟机使用SS-PLS负荷预测的可行性和优势.试验结果表明,基于滑动窗口模型结合SS-PLS,使得预测精度有了显着的改善,即均方根误差为1.77786,平均绝对误差是1.3312,平均绝对误差百分比为0.23836,三者的增量分别5.47%、6.37%、6.12%.该研究可为云平台中虚拟机资源管理和优化提供一种参考方法.
Bibliography:11-2047/S
In order to optimize the infrastructure resource of agricultural cloud video platform efficiently, the virtual machine (VM) placement algorithms need to know the current and future efficiency of VM resource as accurately as possible for potential actions such as service deployment, VM deployment, migration or cancellation. However, there is limited data available for analysis as the samples used in prediction are usually very small. In this paper, a sliding window model was designed to learn from small data sets considering time factor. More importantly, the existing prediction algorithms still have much room to reduce the error. So a sliding window based mathematical method was provided to calculate the aforementioned forecasts, which was combined with PLS and semi-supervised learning (semi-supervised partial least squares, SS-PLS). The feasibility and advantages were analyzed in VM load forecasting with mothod SS-PLS. Compared with auto regression moving average (ARMA), experimental results showed
ISSN:1002-6819
DOI:10.11975/j.issn.1002-6819.2017.z1.034