Research on Coal Quality Using Near-Infrared Spectroscopy Based on Machine Learning

Near-Infrared Spectroscopy (NIRS) is widely used in coal quality analysis due to its advantages of fast detection speed, non-destructive nature, and environmental friendliness. However, traditional spectral analysis methods struggle to achieve high accuracy and robustness when dealing with the complex and nonlinear characteristics of coal compositions. To address these limitations, this study proposes a coal quality analysis method using machine learning models such as Support Vector Machine (SVM), Random Forest (RF), and Deep Neural Networks (DNN). First, the coal's NIRS data is preprocessed and feature extraction is performed using Principal Component Analysis (PCA) and wavelength selection techniques to reduce dimensionality and optimize features. Then, the performance of different machine learning models is compared, and the models' accuracy and stability are optimized through hyperparameter tuning and cross-validation. Experimental results show that the proposed DNN model achieves higher accuracy and better generalization ability in predicting coal components (e.g., moisture, ash, and volatile matter) compared to traditional regression and linear discriminant models. Additionally, a coal quality NIRS analysis system is designed and its effectiveness and feasibility are validated in practical applications. This study provides an efficient and precise technical approach for coal quality analysis, offering valuable insights for quality control in coal production processes.