Document Type : Full Length Research Article
Authors
1
Department of Mechanical Engineering, S N Pate Institute of Technology (SNPIT), Bardoli, Gujarat, India
2
Department of Chemical Engineering,Professor, Sardar Vallabhbhai National Institute of Technology, Surat, Gujarat, India
3
Department of Mechanical Engineering, Retd. Professor, Sardar Vallabhbhai National Institute of Technology, Surat, Gujarat, India
Abstract
India possesses abundant coal reserves, but much of this coal is characterized by high ash content, presenting challenges for efficient utilization. To better understand and model the gasification behavior of high ash content coal, this study presents a novel application of Artificial Neural Networks (ANN) and Machine Learning (ML)-based regression models. Addressing a significant gap in existing literature, the models were developed to predict key output parameters—namely the concentrations of CH4, CO, CO2 and H2 using input features such as elemental composition (C, H, N, S, Ash). Three ML regression techniques—Multiple Linear Regression (MLR), Support Vector Regression (SVR), and Random Forest Regression (RFR)—were employed. Among these, the RFR model exhibited the highest prediction accuracy, with all outputs within permissible error limits and an R2 value of 0.99993. The SVR model also showed satisfactory performance but occasionally produced unrealistic outputs, such as negative CH4 concentrations. MLR, on the other hand, was inadequate for capturing the complex nonlinear relationships in the data. The ANN model achieved strong alignment with experimental results, with minimal average deviations across all target variables, and demonstrated stable performance after 10 training epochs. A comparative analysis between the RF and ANN models suggests that a hybrid modelling approach could further enhance prediction reliability. Overall, the findings confirm the suitability of advanced ML techniques, particularly Random Forest and ANN, for accurately modelling the gasification process of high ash content coal.
Keywords
Main Subjects
)