A comparative study of swarm intelligence and artificial neural networks applications in modeling complex reaction processes
Published in Computer Aided Chemical Engineering, 2022
This study investigated two artificial intelligence techniques, Swarm Intelligence (SI) and Artificial Neural Networks (ANN), aiming to overcome the difficulties of simulating complex processes with unknown reactions and intermediates. These techniques are incorporated in reaction modeling via mass balances and reaction kinetic models. The accuracy and the applicability of the resulting models from ANN and SI were compared in the trained semi-batch reactors and the new continuous flow reactors. The ANN-based model is recommended when the extrapolation is unnecessary, and the data is high in volume and variety at the applied space. In this case, no profound reaction knowledge is required. Otherwise, the SI-based model should be employed, which provides detailed information of the target process and is constrained by physical meaning parameters.
Recommended citation: Wu, M., Di Caprio, U., Elmaz, F., Metten, B., De Clercq, D., Van Der Ha, O., Mercelis, S., Hellinckx, P., Braeken, L., Leblebici, M.E. (2022). A comparative study of swarm intelligence and artificial neural networks applications in modeling complex reaction processes. Computer Aided Chemical Engineering, 51, 175-180