Coal and Gangue Recognition in Fully Mechanized Coal Mining Face using Extreme Boosting Algorithm- Ant Colony Optimization

In recent years, coal became an important source for generating electricity, Industrial processes, steel production etc., as it available extensively and cost-effective than other fossil fuels. These mined coals also contain gangue, which releases harmful gases during burning that resulted in serious effects to environment and living beings if not detected in early stages. The traditional models which Assisted for detection of gangue failed in effectively classifying normal coal and gangue coal. Therefore, this paper proposes Gradient Boosting - Ant Colony Optimization (XGBoost- ACO) method for effectively identifying the gangue coal in fully mechanized coal mining face. The proposed method employed on publicly available Coal Classification dataset, the collected image data is preprocessed under gaussian filter for image screening, crushing and magnetic separation. The preprocessed data features are extracted using Coal Gangue Recognition- Convolutional Neural network (CGR-CNN) recognizing coal and gangue particles to handle unique visual images. Finally, XG-BoostACO is employed for improving the detection accuracy in across different batches of coal and gangue recognition. The proposed XGBoost-ACO method attained high robustness in terms of accuracy, precision, recall and $\mathbf{f 1}$-score and values are (96.45%) of accuracy, (95.50%) of precision, (99.20%) of recall and the f1-score value is $(\mathbf{9 7. 3 7 \%}$) when compared with existing Support Vector Machine (SVM).