Towards Data-Driven NARX ANN Simulation For Optimal Control Of The Flue Gas Desulphurization For Coal Power Plants

Abstract

This paper presents the ANN-based algorithm for data-driven optimal control of desulfurization of the flue gases from Coal Power Plants. We have proposed the NARX recurrent neural network with experimentally selected feedback connection length as the black box model for the first stage of the control process. Then simple brute force algorithm was used to find the optimal level of the reagent added into the system to keep the SOx concentration outlet below the assumed level. This procedure was designed for a known level of SOx concentration inlet. The proposed approach was tested on real data collected from the selected Coal Power Plant in Poland. The simulation that was made confirms that such an approach is effective for coal power plants to increase their energy efficiency and meet the appropriate environmental standards.