Sensor network design based on system-wide reliability criteria. Part II: Formulations and applications. (September 2020)

• Record Type: Journal Article
• Authors: Prakash, Om; Bhushan, Mani; Narasimhan, Sridharakumar; Rengaswamy, Raghunathan
• Is Part Of: Journal of process control. Volume 93(2020)
• Page Start: 14
• Page End: 27
• Publication Date: 2020-09
• Subjects: Sum of disjoint product -- Fault detection and diagnosis -- Estimation of variables -- Tennessee Eastman -- Steam metering; Process control -- Periodicals; Fabrication -- Contrôle -- Périodiques; Process control; Periodicals; Electronic journals; 660.281
• DOI: 10.1016/j.jprocont.2020.07.004
• Languages: English
• ISSNs: 0959-1524

Abstract

Abstract: In part I of this series of articles, the concept of overall system reliability was presented for two applications: reliable estimation of variables for steady state linear flow processes, and reliable fault detection and diagnosis for any process. In this part, systematic generation of the proposed system-wide reliability expression is discussed. In particular, an approach for generating the system reliability using the sum of disjoint product method is presented. This serves as the objective function to be maximized in various constrained optimization formulations for sensor network design, which are also proposed in this work for both applications. A heuristic is proposed to solve the resulting nonlinear integer programming problems. The sensor network design formulations are applied to two benchmark case studies: (1) Tennessee Eastman process, (2) Steam metering process. Results indicate the utility of the proposed sensor network design approach in designing optimal sensor networks that maximize system reliability. Highlights: System reliability objective is efficiently generated for sensor network design. Sum of disjoint product based algorithm is proposed for this efficient generation. Optimization formulations for sensor network design presented. Heuristic to solve the resulting nonlinear integer programming problem proposed. Sensor network design approach applied to two benchmark case studies.