Abstract:
In this paper, we propose a new systematic methodology to identify dynamic changes in the connectivity among various equipment items in a process plant. Drawing analogy t...Show MoreMetadata
Abstract:
In this paper, we propose a new systematic methodology to identify dynamic changes in the connectivity among various equipment items in a process plant. Drawing analogy to neurological systems, the paper develops a framework to obtain dynamic causal model to capture the intrinsic and stimulus-driven connectivity among equipment items. An expectation maximization algorithm following a Bayesian framework is employed to obtain the elements of the connectivity matrices using a maximum likelihood approach. The algorithm is tested on a reactor-separator system. The proposed approach can help to decompose large-scale process plants into strongly-connected and weakly-connected systems. If only strongly-connected equipment items are considered together for control structure selection, this approach can result in a computationally less intensive problem for dynamic controlled variable selection of large-scale process plants.
Published in: 2016 American Control Conference (ACC)
Date of Conference: 06-08 July 2016
Date Added to IEEE Xplore: 01 August 2016
ISBN Information:
Electronic ISSN: 2378-5861