摘要
流程工业系统是一种典型的分布式复杂机电系统,其分布式控制系统(DCS)实时记录着整个系统所有设备的运行信息。为了挖掘出蕴含在DCS监测数据集中的系统运行健康状态信息,全面掌握系统总体运行健康状态,快速准确地辨识和定位故障,提出一种名为状态特征谱的方法,将蕴含在海量高维数据中的系统运行健康状态信息以图像的形式展现在二维平面上。通过观察和分析状态特征谱上像素点的位置和颜色的分布,在企业层面全面掌握系统运行健康状态。分别以化工领域通用的田纳西仿真数据和某化工企业DCS数据集为实例,构造了对应的状态特征谱,验证了所提方法的先进性、创新性和有效性。
Production system of process industry is a typical distributed complex electromechanical system, whose Distributed Control System(DCS) records operational information of all the equipment of the whole system in real time. For the purpose of mining system health status information contained with its DCS data set, fully grasping overall operating health status, quickly and accurately identifying and locating faults, a method named state feature spectrum was proposed, which displayed operating health status concealing inside DCS data set on a two-dimensional plane by the form of digital images. By observing and analyzing position and color distribution of pixels on the spectrum, the plant-wide operating health status could be fully comprehended. DCS data set of Tennessee-Eastman process and a real chemical plant were both used to form their state feature spectrum correspondingly verifying the advancement, innovation and effectiveness of the method.
作者
陈倩
孙锴
CHEN Qian;SUN Kai(School of Mechanical and Eletrical Engineering,Xi'an University of Architecture and Technology,Xian 710055,China;Institute of Microelectronic of Chinese Academy of Science,Beiing 100029,China)
出处
《计算机集成制造系统》
EI
CSCD
北大核心
2022年第8期2430-2439,共10页
Computer Integrated Manufacturing Systems
基金
国家自然科学基金青年科学基金资助项目(51705393)。
关键词
流程工业
分布式控制系统
数据分析
多变量时序数据
状态特征谱
process industry
distributed control system
data analysis
multivariable time series
state feature spectrum
