Automation of data-driven rate equation screening for
heterogeneously-catalyzed reactions
Abstract
Automating the generation of suitable kinetic models could dramatically
improve its application to novel reactions. Therefore, a software tool
was developed to automatically propose rate equations for a catalytic
reaction, purely based on experimental data. The tool screens
theoretical initial rate equations by comparing the trends predicted by
theoretical equations (from a comprehensive library) with those present
in the experimental data, thus eliminating the rate equations that
cannot reproduce the trends in the data. Afterwards, the kinetic models
associated to the feasible rate equations are ranked based on trend
similarity, resulting in an ordered list of models ready for regression.
For most of the tested literature datasets, the tool proposed the same
rate equation as experienced researchers. This is a key first step into
the automation of kinetic modelling that, once generalized, will allow
to its widespread use in the understanding of catalytic reactions.