一种高维数据流在线监测方法  

An OnlineMonitoringMethod for High-Dimensional Data Streams

作　　者：刘先俊 齐德全 施三支[1] LIU Xianjun;QI Dequan;SHI Sanzhi(School of Science,Changchun University of Science and Technology,Changchun,Jilin 130012,China)

机构地区：[1]长春理工大学理学院,吉林长春130012

出　　处：《工业工程与管理》2021年第5期46-50,共5页Industrial Engineering and Management

基　　金：国家自然科学基金(11571191);吉林省自然科学基金(20140101199JC);中央高校基本科研业务费理学院项目(3122017082)。

摘　　要：在线监控高维数据流时,经常遇到部分信息价值较低甚至没有信息价值的数据流对监控结果产生干扰。为了解决这一问题,在统计过程控制中引入了最大信息系数方法,提出了一种基于最大信息系数的高维数据流在线监测方法。该方法能有效筛选出高维数据流中具有较高信息价值的数据流,加快失控状态的报警速度。通过数值模拟和案例分析对所提方法的效果进行检验。结果表明,相较于现有的控制图,所提方法在检测均值漂移时具有更好的监控性能。When online monitoring high-dimensional data streams,it is often encountered that some data streams with low or no information value interferes with the monitoring results.In order to solve this problem,the maximum information coefficient method was introduced into the statistical process control,and a high-dimensional data streams online monitoring method based on the maximum information coefficient was proposed.This method could effectively filter out the data streams with high information value in high-dimensional data streams,and accelerate the alarm speed of out-ofcontrol state.The effectiveness of the proposed method was tested by numerical simulation and case analysis.The results show that compared with the existing control chart,the proposed method has better monitoring performance in detecting the mean shift.

关 键 词：最大互信息系数 相关程度 多元累积和控制图 高维数据流 

分 类 号：O213.1[理学—概率论与数理统计]