摘要
为了对大数据进行分流处理,需要对大数据分流系统设计方法进行研究。采用当前分流系统设计方法对大数据进行分流处理时,存在聚类准确性低和分流效率低的问题。在优化聚类算法的基础上提出一种大数据分流系统设计方法,采用空间重构方法对分流系统中的大数据进行非线性映射处理,得到数据的时间序列信息模型,通过指标数据投影方法得到数据在信息模型中的高维映射矢量,根据数据的高维映射矢量构建搜索目标函数,并求得目标函数的极值,利用极值计算大数据的时延尺度特征值,根据数据时延尺度特征值采用粒子群算法对大数据进行聚类处理,根据聚类结果对大数据进行分流处理,完成大数据分流系统的设计。仿真结果表明,所提方法的特征提取结果精准度高,系统分流所用时间少,验证所提方法的聚类准确性高、分流效率高。
In order to shunt big data,we need to research the design method of big data distribution system. Based on the optimization clustering algorithm,a method to design big data distribution system was proposed.Firstly, the spatial reconstruction method was used to perform nonlinear mapping on big data in the distribution system,and then time series information model of data was obtained.Secondly,the index data projection method was used to get high-dimensional mapping vector of data in information model.Thirdly,high-dimensional mapping vector of data was used to construct the search target function,and thus to obtain the extremum of objective function.Moreover,the extremum was used to calculate the time-delay scale eigenvalue of big data.According to the time-delay scale ei-genvalue,the particle swarm algorithm was used to cluster big data.Finally,the big data were distributed according to the clustering results.Thus,the design of big data distribution system was completed.Simulation results show that the proposed method has higher accuracy in feature extraction and less time for system distributary.Meanwhile,the proposed method has high clustering accuracy and high distributary efficiency.
作者
张晓婷
李茵
唐晶磊
ZHANG Xiao-ting;LI Yin;TANG Jing-lei(College of Information Engineering,Northwest Agricuhural and Forest University,Yangling Shanxi 712100,China)
出处
《计算机仿真》
北大核心
2018年第12期204-207,共4页
Computer Simulation
基金
融入直觉化交互机理的人机界面全息设计与评价理论研究(61472314)
关键词
优化聚类算法
大数据
分流系统
Optimization clustering algorithm
Big Data
Separate system
