Litcius/Paper detail

Chatter Detection and Diagnosis in Hot Strip Mill Process With a Frequency-Based Chatter Index and Modified Independent Component Analysis

Ha-Nui Jo, Byeong Eon Park, Yumi Ji, Dong-Kuk Kim, Jeong Eun Yang, In−Beum Lee

2020IEEE Transactions on Industrial Informatics30 citationsDOI

Abstract

In this article, we propose a framework to monitor the chatter phenomenon and to diagnose the cause variables of chatter occurred in the hot strip mill process (HSMP). For monitoring chatter, we develop a chatter index (CI) that quantifies chatter to confirm its occurrence. Based on the data classified as normal by the CI, a multivariate statistical process monitoring model for detecting chatter is constructed using the modified independent component analysis (MICA) method. The monitoring results show that the model based on the MICA outperforms other models based on the principal component analysis and independent component analysis. For the diagnosis of the cause variables of detected chatter, various contribution plots can be used. In this article, we develop a relative contribution plot for a more obvious diagnosis than the existing contribution plot. Using this, we diagnose and analyze the cause variables of the detected chatter in the HSMP.

Topics & Concepts

Principal component analysisComponent (thermodynamics)Process (computing)Multivariate statisticsIndependent component analysisPlot (graphics)StatisticsMathematicsComputer scienceEngineeringPattern recognition (psychology)Data miningArtificial intelligenceOperating systemThermodynamicsPhysicsFault Detection and Control SystemsMineral Processing and GrindingAdvanced Statistical Process Monitoring