基于时序关联智慧城市边缘数据异常检测算法  

Outlier data detection algorithm at edge of smart city based on time series correlation

作　　者：谢小川 周绍军[2] 黎力 黎明[4] XIE Xiao-chuan;ZHOU Shao-jun;LI Li;LI Ming(College of Teacher Education,Sichuan Normal University,Chengdu 610068,China;Department of Information Engineering,Sichuan Water Conservancy Vocational College,Chongzhou 611231,China;School Software Engineering,Tongji University,Shanghai 201804,China;College of Computer Science,Sichuan Normal University,Chengdu 610101,China)

机构地区：[1]四川师范大学教师教育学院,四川成都610068 [2]四川水利职业技术学院信息工程系,四川崇州611231 [3]同济大学软件学院,上海201804 [4]四川师范大学计算机科学学院,四川成都610101

出　　处：《计算机工程与设计》2022年第7期1879-1889,共11页Computer Engineering and Design

基　　金：国家自然科学基金项目(61701331)。

摘　　要：针对智慧城市边缘感知数据类型多、数据维度大和存在数据异常等问题,提出基于时序的边缘检测异常数据算法。对解决该问题的基于边缘计算的智慧城市物联网、大数据分析框架进行设计,同时设计边缘服务增强现实框架;对智慧城市边缘检测异常数据问题进行定义,设计检测流程和时序关联计算算法,提出基于时序关联的智慧城市边缘检测异常数据算法。对设计的算法,利用感知设备采集数据,进行大量实验与仿真对比分析,实验结果表明,该算法在解决时序关联多维数据异常检测的准确率和召回率方面,具有一定的优越性。According to the problems of multiple types of data,large data dimensions and data anomalies in edge perception of smart city,a time-based edge detection algorithm for outlier data was proposed.The edge-computing-based IoT of smart city and big data analysis framework were designed to solve these problems,and the augmented reality framework of edge service was designed at the same time.The problem of outlier data in edge detection of smart city was defined,a calculating algorithm of detection process and temporal association was designed,and an algorithm for detecting outlier data in edge detection of smart city based on temporal association was proposed.Sensing devices were used to collect data,a large number of experiments and simulation comparisons were carried out for the designed algorithm.Results show that the algorithms have certain advantages in accuracy and recall rate of temporal-associated multi-dimensional outlier data detection.

关 键 词：数据异常 检测算法 时序 边缘计算 智慧城市 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]