A novel detection of correlated alarms with delays based on improved block matching similarities. (March 2020)

• Record Type: Journal Article
• Authors: Yang, Bo; Wang, Hai-run; Li, Hong-guang; He, Ya-dong
• Is Part Of: ISA transactions. Volume 98(2020)
• Page Start: 393
• Page End: 402
• Publication Date: 2020-03
• Subjects: Alarm sequence -- Time series analysis -- Correlation analysis -- Block matching -- Time delay analysis; Engineering instruments -- Periodicals; Engineering instruments; Periodicals; Electronic journals; 629.805
• DOI: 10.1016/j.isatra.2019.07.011
• Languages: English
• ISSNs: 0019-0578

Abstract

Abstract: Statistical analysis method has emerged as a general approach to detect relation alarms in the process industries. However, the delay between related alarms is the main cause leading to wrong analysis results from traditional approaches to detect correlated alarms. This paper proposed a novel detection of correlated alarms based on block matching similarities with delay (BMS-d). First, blocking alarm data sequence method is to transform alarm data into time node sequences, which is able to reduce the calculation burden of the correlation analysis. Second, a novel maximal block correlation coefficient method is presented to estimate the correlation delay between alarms. Third, a novel method is proposed to detect correlated alarms based on the block matching similarities and related alarm delay information. A numerical case and TE process are employed to demonstrate the effectiveness and efficiency of the proposed method. Highlights: This paper proposed A Novel Detection of Correlated based on improved BMS. The proposed BMS-d method aims at reducing the calculation burden of the correlation analysis. The proposed BMS-d method can estimate transfer time delay between alarm variables.