AI-Based Optimization of Excess Air Supply for Minimizing Unburnt Carbon and Thermal Losses in Coal-Fired Thermal Power Plants

Abstract

Thermal power plants operating on coal remain a very important contribution to the generation of electricity but the burning deficiencies like carbon loss that is not burned and heavy loss of flue gas shows a serious impact on the efficiency of the plant. The excess air supply is a very important combustion parameter and it has a direct effect on the completeness of the combustion and the losses in terms of thermodynamics which presents a complicated trade-off. In this paper, a multi-objective optimization model using artificial intelligence (AI) is suggested to determine the best excess air level which has minimal unburnt carbon and flue gases losses and a maximum boiler efficiency. A Genetic Algorithm (GA) was used as an internal optimizer with industrial operating conditions, and a hybrid AI model based on Long Short-Term Memory (LSTM) and XGBoost regression was created to predict the power output and boiler efficiency with high accuracy. Simulation made in MATLAB was done with the boiler predicted data-sheet aligned parameters so that it is industrial relevant. The findings show that the more excess air, the lower the amount of unburnt carbon in the bottom ash to about 2 per cent, at 15 per cent excess air, thus confirming the presence of almost complete combustion. Nevertheless, at this level, the increase of dry flue gas heat loss is almost 7%, which is the reason why optimization is needed.