基于灰色关联分析法的物联网数据异常检测方法  

Anomaly Detection Method of Internet of Things Data Based on Grey Correlation Analysis Method

作　　者：陈浩杰 CHEN Haojie(Zhejiang Post and Telecommunication College,Shaoxing,Zhejiang 312366,China)

机构地区：[1]浙江邮电职业技术学院,浙江绍兴312366

出　　处：《移动信息》2024年第5期214-216,共3页MOBILE INFORMATION

摘　　要：物联网数据具有构成复杂、规模庞大、异常形式多样化的特征,导致基于传统异常检测方法得到的结果正确率偏低。文中提出了一种基于灰色关联分析法的物联网数据异常检测方法。该方法利用灰色关联模型计算得到灰色关联系数,对原始物联网数据中的自有变量相关性进行排序,随后采用前向选择法进行变量排序选择,确定反推得到的特征参数误差达到最小值时的变量,并将其作为最优变量子集;在异常检测阶段,用最优变量子集对物联网数据进行异常位置寻优,实现异常检测。测试结果表明,该方法在不同攻击下,异常数据检测结果的F1值未出现明显的波动,对应的F1值为0.84~0.90。The Internet of Things data has the characteristics of complex composition,large scale and diverse anomaly forms,resulting in low accuracy of results obtained based on traditional anomaly detection methods.This paper proposes a method for anomaly detection of Internet of Things data based on grey correlation analysis.The method uses the grey correlation coefficient calculated by the grey correlation model to sort the correlation of its own variables in the original Internet of Things data,and then uses the forward selection method to select the variables in order to determine the variables when the error of the characteristic parameters obtained by inversion reaches the minimum value,and uses it as the optimal variable subset;in the anomaly detection stage,the optimal variable subset is used to optimize the anomaly position of the Internet of Things data to achieve anomaly detection.The test results show that the F1 value of the abnormal data detection results of this method does not fluctuate significantly under different attacks,and the corresponding F1 value is 0.84~0.90.

关 键 词：灰色关联分析法 物联网数据 异常检测 灰色关联模型 灰色关联系数 自有变量相关性 前向选择法 最优变量子集 

分 类 号：TP212[自动化与计算机技术—检测技术与自动化装置]