Executive Summary : | The "Gross Calorific Value" (GCV), a measure of a coal's heat content, is one of the most crucial quality parameters since it affects the price of coal and its use in industrial sectors. The standard approach is labor and time-intensive but effective, whereas unconventional approaches are usually unutilized due to differences among researchers regarding their efficacy. There are several ways to predict GCV, including utilizing the data from the proximate analysis, ultimate analysis, or both. Further, ash and moisture are both known to have a significant impact on GCV estimation. The aim of this research is to use machine learning algorithms to take advantage of the natural variability in the mid-infrared spectral response of coal samples to determine the heat content in coal. The range of coal with varied carbon, hydrogen, nitrogen, and oxygen content as well as impurities in the form of ash, is covered by the spectrum response because it considers the effects of both organic and inorganic matter in coal. Also, using mathematical models, the applicability of data from the proximate and ultimate analyses, with different ash contents, will be investigated for the determination of GCV. The modeling and its validation will be done using a training and testing set by employing K-fold cross-validation so that the test set is independent of the training set. In order to compare and validate model predictions, the GCV obtained using the bomb calorimeter in accordance with ASTM guidelines will be used as a reference. The standard two-tailed t-test and F-test for mean and variance will be carried out to check for any significant difference between the pair of values of observed GCV (GCVlab, wt.%) using bomb calorimeter data and the model predicted GCV (GCVFTIR, wt.%) using FTIR data. In initial research using a very small set of samples, the prediction model using data from mid-infrared FTIR spectroscopy and the machine learning technique using loss function performs well for coal samples, suggesting the need for future development and extensive testing. The suggested approach based on FTIR data can be a useful tool for examining GCV in ash-rich coals in India and other countries. |