面向设备群体的工况数据异常检测方法

doi:10.13196/j.cims.2013.12.yaoxinxin.2993.9.2013129

摘要/Abstract

摘要： 为了全面便捷地检测出有效的潜在的设备异常,基于设备大量有价值的工况数据,对设备异常检测方法进行研究,提出一种面向设备群体的工况数据异常检测方法。通过对设备状态监测数据潜在异常特征进行分析挖掘,识别设备运行中的异常状态和用户异常行为,提高产品服务质量。给出了设备异常检测原理,对数据预处理、特征提取、正常特征空间分布建模、异常检测和人工验证过程进行了详细阐述。通过对5000多万条真实监测数据进行异常检测实验,验证了所提方法的可行性和有效性。

关键词: 设备群体, 工况数据, 特征空间, 分布建模, 异常检测

Abstract: To detect effective potential equipment abnormal comprehensively,the detecting method based on amount of valuable equipment condition monitoring data was researched,and a device group-oriented method for abnormal floor data detecting was proposed.Through analyzing and mining the potential abnormal features of equipment condition monitoring data,the abnormal state and the user abnormal behavior in equipment operation were identified,and the service quality of products was improved.The principle of abnormal detecting was presented,and the processes of data preprocessing,feature extraction,spatial distribution of normal features modeling,abnormal detecting and artificial validation were elaborated in detail.More than 50 million real monitoring data were detected in anomaly detecting experiment,and the feasibility and effectiveness of the method were validated.

Key words: device group, floor data, feature space, distribution modeling, abnormal detecting

中图分类号:

TP391
TP319

姚欣歆,刘英博,赵炯,胡游乐,张力. 面向设备群体的工况数据异常检测方法[J]. , DOI: 10.13196/j.cims.2013.12.yaoxinxin.2993.9.2013129.

YAO Xin-xin,LIU Ying-bo,ZHAO Jiong,HU You-le,ZHANG Li. Device group-oriented method for abnormal floor data detecting[J]. , DOI: 10.13196/j.cims.2013.12.yaoxinxin.2993.9.2013129.

参考文献

［1］MIN Rui,XU Liming,WEI Jujun,et al.Anomaly detection of running state of machining system［J］.Chinese Journal of Scientific Instrument,2006,27(6):1777-1778 (in Chinese).［闵睿,许黎明,魏巨隽,等．机械加工系统运行状态异常检测方法的研究［J］．仪器仪表学报,2006,27(6):1777-1778.］
[2]XIAO Gui.Anlmaly detection methods and measures for operating state of mechanical processing system［J］.Technological Development of Enterprise,2011,30(3):83-84(in Chinese).［肖贵.机械加工系统运行状态异常检测方法及措施探讨［J］.企业技术开发,2011,30(3):83-84.］
[3]MCARTHUR S D J,BOOTH C D,MCDONALD J R,et al.An Agent-based anomaly detection architecture for condition monitoring［J］.IEEE Transactions on Power Systems,2005,20(4):1675-1682.
[4]ZHANG Hongli,LIU Shulin,JIAO Wenhui,et al.Equipment abnormal degree detection approach based on variable threshold information detector［J］.Journal of Mechanical Engineering,2013,49(8):25-30(in Chinese).［张宏利,刘树林,缴文会,等.基于可变阈值信息检测器的设备异常度检测方法［J］.机械工程学报,2013,49(8):25-30.］
[5]NI Jingfeng,LIU Lihua,GU Yujiong.Outliers detection in time series of measured data based on least square support vector machine algorithm［J］.Journal of North China Electric Power University,2008,35(3):62-66(in Chinese).［倪景峰,刘丽华,顾煜炯.基于最小二乘支持向量机算法的测量数据时序异常检测方法［J］.华北电力大学学报,2008,35(3):62-66.］
[6]WANG Tianyang,CHENG Weidong,LI Jianyong,et al.Anomaly detection for equipment condition via cross-correlation approximate entropy［C］//Proceedings of 2011 International Conference on Management Science and Industrial Engineering. Washington,D.C.,USA:IEEE,2011:52-55.
[7]GAO Yu,YANG Tianshe,XU Minqiang,et al.An unsupervised anomaly detection approach for spacecraft based on normal behavior clustering［C］//Proceedings of the 5th International Conference on Intelligent Computation Technology and Automation.Washington,D.C.,USA:IEEE,2012:478-481.
[8]FEI Heliang,XU Xiaoling,LU Xiangwei.Masking effect on tests for outliers［J］.Chinese Journal of Applied Probability and Statistics,2002,18(2):141-151(in Chinese).［费鹤良,徐晓岭,陆向薇.异常数据检验的屏蔽效应［J］.应用概率统计,2002,18(2):141-151.］
[9]LIU Mingji,WANG Xiufeng,HUANG Yalou.Data preprocessing in data mining［J］.Computer Science,2000,27(4):54-57(in Chinese) .［刘明吉,王秀蜂,黄亚楼.数据挖掘中的数据预处理［J］.计算机科学,2000,27(4):54-57.］
[10]WANG Daling,YU Ge,BAO Yubin,et al.A representation about domain knowledge for reprocesses of data mining［J］.Mini-Micro Systems,2003,24(5):863-868 (in Chinese).［王大玲,于戈,鲍玉斌,等.一种面向数据挖掘预处理过程的领域知识的分类及表示［J］.小型微型计算机系统,2003,24(5):863-868.］

[1]	孙笑笑,侯文杰,沈沪军,应钰柯,俞东进. 基于上下文感知的多角度业务流程在线异常检测方法[J]. 计算机集成制造系统, 2021, 27(9): 2532-2541.
[2]	丁汀,颜登程,张以文,周珊. 融合煤矿多维时序数据的瓦斯异常检测算法[J]. 计算机集成制造系统, 2020, 26(6): 1651-1659.
[3]	魏懿,曹健. 基于机器学习的流程异常预测方法[J]. 计算机集成制造系统, 2019, 25(第4): 864-872.
[4]	赵卫东,刘海涛. 流程挖掘在流程优化中的应用[J]. 计算机集成制造系统, 2014, 20(10): 2632-2642.