Preprint has been published in a journal as an article
DOI of the published article https://doi.org/10.1016/j.compchemeng.2023.108319
Preprint / Version 1

Hybrid modelling of a batch separation process

##article.authors##

  • Ulderico Di Caprio Center for Industrial Process Technology, Department of Chemical Engineering, KU Leuven, Agoralaan Building B, 3590 Diepenbeek, Belgium https://orcid.org/0000-0001-5194-8721
  • Min Wu Center for Industrial Process Technology, Department of Chemical Engineering, KU Leuven, Agoralaan Building B, 3590 Diepenbeek, Belgium
  • Furkan Elmaz IDLab, Faculty of Applied Engineering, University of Antwerp-imec, Sint-Pietersvliet 7, 2000 Antwerp, Belgium
  • Yentl Wouters Janssen Pharmaceutica N.V., Chemical Process R&D, Turnhoutseweg 30, B-2340 Beerse, Belgium
  • Niels Vandervorst Janssen Pharmaceutica N.V., Chemical Process R&D, Turnhoutseweg 30, B-2340 Beerse, Belgium
  • Ali Anwar IDLab, Faculty of Applied Engineering, University of Antwerp-imec, Sint-Pietersvliet 7, 2000 Antwerp, Belgium https://orcid.org/0000-0002-5523-0634
  • Siegfried Mercelis IDLab, Faculty of Applied Engineering, University of Antwerp-imec, Sint-Pietersvliet 7, 2000 Antwerp, Belgium https://orcid.org/0000-0001-9355-6566
  • Steffen Waldherr KU Leuven, Department of Chemical Engineering, Celestijnenlaan 200F-bus 2424, Leuven 3001, Belgium https://orcid.org/0000-0002-0936-579X
  • Peter Hellinckx Faculty of Applied Engineering, University of Antwerp, Groenenborgerlaan 171, 2000 Antwerp, Belgium
  • M. Enis Leblebici Center for Industrial Process Technology, Department of Chemical Engineering, KU Leuven, Agoralaan Building B, 3590 Diepenbeek, Belgium https://orcid.org/0000-0003-4599-9412

DOI:

https://doi.org/10.31224/2874

Keywords:

hybrid modelling, dynamic system, solvent switch, optimization, statistical modelling, data value

Abstract

Applying machine learning (ML) techniques is a complex task when the data quality is poor. Integrating first-principle models and ML techniques, namely hybrid modelling significantly supports this task. This paper introduces a novel approach to developing a hybrid model for dynamic chemical systems. The case in analysis employs one first-principle structure and two ML-based predictors. Two training approaches (serial and parallel), two optimisers (particle swarm optimisation and differential evolution) and two ML functions (multivariate rational function and polynomial) are tested. The polynomial function trained with the differential evolution showed the most accurate and robust results. The training approach does not significantly affect the hybrid model accuracy. However, the main effect of the training approach is on the robustness of the parameter predictions. The coefficients of determination (R2) on the test batches are above 0.95. In addition, it showed satisfactory extrapolation capabilities on different production scales with R2>0.9.

Downloads

Download data is not yet available.

Downloads

Additional Files

Posted

2023-03-14