基于大数据分析技术的云计算资源预测研究  

Research on Cloud Computing Resource Prediction Based on Big Data Analysis Technology

作　　者：卢思安[1] 侯国庆[2] LU Si-an;HOU Guo-qing(College of Computer and Information Engineering,Inner Mongolia Agricultural University,Hohot Inner Mongolia 010018,China;College of Economics and Management,Inner Mongolia Agricultural University,Hohhot Inner Mongolia 010018,China)

机构地区：[1]内蒙古农业大学计算机与信息工程学院,内蒙古呼和浩特010018 [2]内蒙古农业大学经济管理学院,内蒙古呼和浩特010018

出　　处：《计算机仿真》2022年第10期502-505,537,共5页Computer Simulation

摘　　要：为有效呈现云计算资源负载的动态波动情况,研究基于大数据分析技术的云计算资源预测方法。在海量云计算资源负载的时间序列数据内任意选取一个时间序列数据;采用时间序列混沌分析方法预处理所选时间序列数据,通过互相关方法确定时间序列相空间重构的最佳嵌入维和延迟时间;依照最佳嵌入维和延迟时间构建一个多维时间序列,将多维时间序列作为云计算资源负载预测建模的学习样本;采用支持向量机训练学习样本,构建云计算资源负载预测模型,采用蝙蝠算法优化预测模型内惩罚因子和径向基核函数宽度参数,完成海量云计算资源负载时间序列数据预测。测试结果显示所研究方法具有较好的泛化能力,单步预测条件下能够获取高精度的测试对象预测结果,多步预测条件下依旧能够较好的呈现测试对象的波动情况。In order to effectively present the dynamic fluctuation of cloud computing resource load, the cloud computing resource prediction method based on big data analysis technology was studied in the paper.Time series data were randomly selected from the time series data loaded by massive cloud computing resources and preprocessed by time series chaos analysis method. Then the best embedding dimension and delay time of time series phase space reconstruction were determined by cross-correlation method. A multi-dimensional time series was constructed according to the best embedding dimension and delay time, and the multi-dimensional time series was used as the learning sample of cloud computing resource load forecasting modeling. Support vector machine(SVM) was used to train learning samples to build a cloud computing resource load prediction model. Bat algorithm was used to optimize the penalty factor and radial basis function width parameters in the prediction model to complete the massive cloud computing resource load. The test results show that the proposed method has good generalization ability, can obtain high-precision test object prediction results under the condition of one-step prediction, and still can better present the fluctuation of test object under the condition of multi-step prediction.

关 键 词：大数据分析 云计算资源 负载预测 混沌时间序列 互相关 支持向量机 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]