Study of the Current-Voltage Characteristics of Membrane Systems Using Neural Networks

The aim of the investigation is the mathematical modeling of processes that take place in electro membrane systems (EMS). The article is devoted to the use of machine learning and artificial intelligence methods for a study of the current-voltage characteristic (CVC) of membrane systems, taking into account the space charge, electroconvection, forced convection, non-catalytic reaction of dissociation/recombination of water molecules, as well as taking into account the presence and absence of spacers. The creation of neural networks allowed to simulate the behavior of CVC for a wider range of input parameter values than with the help of currently available numerical algorithms and significantly to save time and computing resources. During the work, several different neural network architectures were developed and tested and the best neural network was selected for modeling CVC behavior, including convolutional and LSTM (Long Short-Term Memory) layers with the best predictive results on test samples.