Integration of Adaptive Neuro‐Fuzzy Modeling for Online Coal Quality Assessment in Coal‐Fired Power Plants

South Africa continues to rely on coal‐fired power plants for energy. The transition to renewable sources has begun, but this does not limit research on salvaging coal‐fired power plants. Considering the coal reserves in South Africa, before we fully give up on coal‐fired power plants due to their greenhouse gas emission concerns, research can continue to improve coal‐fired power plant performance. This study focused on developing a correlation that can be used to estimate the gross calorific value (GCV) of coal. The GCV of coal is an important metric in coal‐fired power plants (CFPP), since it shows the energy content of coal and can be used to analyze boiler performance and efficiency. Previous research has involved extensive work on developing correlations that can be used to estimate the GCV of coal using the ultimate and proximate analysis of coal, and it was found that these methods lead to delays in estimating the GCV of coal. This study investigated estimating the GCV of coal using live plant data, specifically the flue gas constituents, focusing on O ₂ , C O ₂ , N ₂ and H ₂ O . These flue constituents’ mass fractions have been confirmed to be measured at the CFPP. The methodology used included a mathematical simulation of coal combustion, which was validated using the literature; the model was then run using coal sample data from various coal grades and geographic regions. An adaptive network‐based fuzzy inference system (ANFIS) model was trained and validated to estimate the GCV of coal, and the model provided a coefficient of determination ( R ² ) of 0.9107, RMSE of 1.661 MJ/kg, MAE of 1.189 MJ/kg, and a MAPE of 4.79%, which was favorable as compared to a multiple linear regression model. The ANFIS model showed strong predictive capability and adaptability, making it a promising tool for future energy and combustion‐related applications.