Observability and redundancy classification in multicomponent process networks

Abstract

Classification of variables and measurements in multicomponent flow networks is treated in this paper. Classification rules are derived that exploit the relationship between cutsets of the process graph and certain graphs derived from it, and the solvability of the relevant equations. These rules are incorporated in two graph-oriented algorithms for observability and redundancy classification.

Extent computation under rank-deficient conditions

“…These concentrations are labeled as structurally unobservable, similarly to Kretsovalis and Mah (1987). Among the remaining unlabeled species, one then finds those species whose columns in N o contain only known stoichiometric parameters.…”

Instrumentation in Processes and Automation

“…The effect of data reconciliation and gross error detection in parameter estimation has been analyzed by MacDonald and Howat [52] as well as by Serth et al [87] and Pages et al [73]. Among the approaches based on linear algebra, Kretsovalis and Mah [44] proposed a combinatorial search based on the effect of the variance of measurements on the precision of reconciled values. Tjoa and Biegler [91] explored methods for the estimation of parameters in differential-algebraic equation systems.…”

CGD-GA: A graph-based genetic algorithm for sensor network design

“…A comprehensive review of the various SND strategies can be consulted in [3]. Kretsovalis and Mah [14] developed methods based on linear algebra in order to design sensor networks and quantify the effects of estimated variables for mass-flow processes. In turn, Maquin et al [16] and Madron and Veverka [15] used the same approach, but aiming at a minimum cost configuration.…”