Integration of neural network-based prediction for enhanced process controllability: application in hydrogen iodide decomposer for hydrogen production via water splitting process

Abstract

Sulfur-Iodine Thermochemical cycle (SITC) process can be broken down into 3 sections. Based on the literature, there is limited study on the third section which is hydrogen iodide (HI) decomposition. In this work, a study to develop a model and controller of the HI decomposition is carried out. The goal of this work is to address this important gap, and more specifically to focus on the controllability of the HI decomposition section. Before the control and simulation study can be performed, a dynamic model of a HI decomposer is first established to obtain the baseline for the necessary data to be fed into the Artificial Neural Network for training to predict the correct outcome accordingly. The proposed controller is a Multi-Scale Control (MSC) integrated into an Artificial Neural Network (ANN) model (ANN-MSC-based-PID). It is worth highlighting that; the proposed model-based control strategy has proven to effectively control the HI decomposition reactor.