The effectivity analysis of adding sensors for improving model based fault isolability properties. (October 2018)

• Record Type: Journal Article
• Authors: Chen, Changxiong; Chen, Liping; Ding, Jianwan; Wu, Yizhong
• Is Part Of: Journal of process control. Volume 70(2018)
• Page Start: 123
• Page End: 132
• Publication Date: 2018-10
• Subjects: Model based fault diagnosis -- Structural analysis -- Sensor placement -- Fault isolability; Process control -- Periodicals; Fabrication -- Contrôle -- Périodiques; Process control; Periodicals; Electronic journals; 660.281
• DOI: 10.1016/j.jprocont.2018.07.007
• Languages: English
• ISSNs: 0959-1524

Abstract

Highlights: Using structural analysis the effectivity of adding sensors for enhancing fault isolability properties is presented. The condition of improving fault isolability properties is provided using the correlations between fault equations. Through structural model decomposition, all the model equations and state variables of a diagnosis system are classified. All the state variables are split into two types of variable sets, which reduce the selection space of additional sensors. Abstract: In order to improve the fault isolability properties of a diagnosis system, a usual way is to add sensors for measuring extra state variables. Unlike previous literatures that usually focus on adding new sensors to fulfill diagnosis specifications, this paper aims to analyze the influences on the isolability properties when more variables are measured. Using structural model decomposition, the equations and variables of a system model are separated into different categories. Then, the association relations between the fault equations can be extracted, which imply the fault isolability. According to the partitions of variables, the selection space of adding sensors is significantly reduced. It is useful to optimize sensor placement and enhance fault isolability. Moreover, our approach need not build a fault signature matrix for fault diagnosis. The efficient algorithms based on this approach are proposed and the simulation of a four-tank system indicates the validity of the method.