MWCVA方法的带钢热连轧过程故障检测研究  

Research on Fault Detection of Hot Strip Rolling Process BasedonMWCVAMethod

作　　者：张瑞成 李禹亭 梁卫征 周亚罗 ZHANG Rui-cheng;LI Yu-ting;LIANG Wei-zheng;ZHOU Ya-luo(College of Electrical Engineering,North China University of Science and Technology,Hebei Tangshan 063210,China)

机构地区：[1]华北理工大学电气工程学院,河北唐山063210

出　　处：《机械设计与制造》2023年第7期65-68,共4页Machinery Design & Manufacture

基　　金：河北省自然科学基金资助项目(F2018209201)。

摘　　要：带钢热连轧过程中采集的数据具有较强的自相关性、互相关性以及时变性,典型变量分析方法(CVA)虽然能够解决自相关性和互相关性的问题,但是由于过程数据存在时变性,导致以前构建的监控模型不再适用于现在的数据,因此提出了一种基于滑动窗口的典型变量分析方法(MWCVA)。该方法首先通过初始窗口构建CVA模型和统计量,解除了数据之间的自相关性和互相关性,然后通过滑动窗口更新过程数据,不断更新CVA模型和统计量,解决了时变性导致检测结果不准确的问题,最后通过控制限判断是否发生故障,在监测系统内部状态空间的同时监测外部状态空间的变化,更加全面地进行故障检测。通过带钢热连轧过程(HSMP)案例的仿真研究,对比CVA、MWCVA的检测效果,证明了MWCVA对故障识别的精度高达100%,误报率不足0.5%。The data collected in the hot strip mill process has strong autocorrelation,cross-correlation and time-varying.The canonical variable analysis(CVA)can solve the problem of autocorrelation and cross-correlation.However,due to the time-varying of process data,the previous monitoring model is no longer applicable to the current data.Therefore,a canonical variable analysis based on moving window is proposed named MWCVA.Firstly,the CVA model and statistics are constructed through the initial window to remove the autocorrelation and cross-correlation between the data.Then,the process data is updated by moving window,and the CVA model and statistics are continuously updated to solve the problem of inaccurate detection results caused by time-varying.Finally,the fault is judged by the control limit.Through the simulation study of hot strip mill process case,the detection effect of CVA and MWCVA is compared.It is proved that the accuracy of MWCVA for fault identification is as high as 100%,and the false alarm rate is less than 0.5%.

关 键 词：故障检测 时变性 滑动窗口 典型变量分析 带钢热连轧 

分 类 号：TH16[机械工程—机械制造及自动化] TP27[自动化与计算机技术—检测技术与自动化装置]