作者: Hamza Albazzaz , Xue Z. Wang , Fatma Marhoon
DOI: 10.1016/J.JPROCONT.2004.06.007
关键词:
摘要: This paper describes a comparative study of multidimensional visualisation technique and multivariate statistical process control (MSPC) for historical data analysis. The uses parallel coordinates which visualise using two dimensional presentations allow identification clusters outliers, therefore, can be used to detect abnormal events. is based on database covering 527 days operation an industrial wastewater treatment plant. It was found that both the MSPC T2 chart captured same 17 as “clearly abnormal” another eight “likely abnormal”. Pattern recognition K-means clustering also applied in literature have identified 14 out days.