Toggle Main Menu Toggle Search

ePrints

S-systems and evolutionary algorithms for the inference of chemical reaction networks from fed-batch reactor experiments

Lookup NU author(s): Dr Dominic Searson, Dr Mark Willis, Professor Allen Wright

Downloads

Full text for this publication is not currently held within this repository. Alternative links are provided below where available.


Abstract

This article aims to demonstrate the potential of the S-system methodology to construct hybrid models of fed-batch reaction systems in which both the reaction network equations and kinetic parameters are unknown. Furthermore, the article aims to show that the hybrid S-systems models can be used as an aid in the inference of the structure of unknown reaction networks from simple fed-batch experimental data and very limited a priori knowledge of the products and reactants. The concept of S-systems is that of a power law formalism for the canonical approximate representation of dynamic non-linear systems. This formalism, which has origins in biochemical systems theory (BST), has the useful property that the causal structure of a network is dictated by the values of its parameters and so network inference problems (e.g. elucidating the topology of a network of reactive species) can be recast as parameter estimation problems. This methodology has been applied in a number of biological fields including metabolic pathway analysis and the inference of gene regulatory networks. Here, the methodology is adapted as a hybrid modelling tool to aid the structural identification of chemical reaction networks in fed-batch reactors. The principle of the approach is demonstrated with simulated noisy data from fed-batch experiments using a reaction network consisting of 5 chemical species and 4 elementary reactions. A co-evolutionary algorithm (comprising binary and real valued components) is used to evolve the topology and the parameter values of the S-system equations concurrently. Unlike many other hybrid modelling approaches, these S-system equations can then be interpreted in order to construct a cause and effect network diagram that accurately reflects the underlying chemical reaction network.


Publication metadata

Author(s): Searson, D. P., Willis, M. J., Home, S., Wright, A. R.

Publication type: Conference Proceedings (inc. Abstract)

Publication status: Published

Conference Name: CHISA 2006: 17th International Congress of Chemical and Process Engineering

Year of Conference: 2006

Publisher: Czech Society of Chemical Engineering

Library holdings: Search Newcastle University Library for this item

ISBN: 8086059456


Actions

Link to this publication


Share