Gradient Boosting Regression Tree Optimized with Slime Mould Algorithm to Predict the Higher Heating Value of Municipal Solid Waste

Abstract

The production of municipal solid waste (MSW) has led to an unprecedented level of environmental pollution, worsening the global challenges posed by climate change. Researchers and policymakers have recently made significant strides in the field of sustainable and renewable energy sources, which are viable from technological, environmental, and economic perspectives. Consequently, the waste-to-energy programs enhance nations’ socioeconomic status while positively impacting the environment. To predict the higher heating value (HHV) of MSW fuel based on carbon, hydrogen, oxygen, nitrogen, and sulfur content, the current study introduces a Gradient Boosting Regression Tree (GBRT) model optimized with the Slime Mold Algorithm (SMA). This model was evaluated using an additional 50 data points after being trained with 202 MSW biomass data points. The performance of the model was assessed using three metrics: root mean square error (RMSE), mean absolute error (MAE), and the coefficient of determination (R2). The results indicated that our model outperformed previously developed models in terms of accuracy and reliability. Additionally, a graphical user interface (GUI) was developed to facilitate the practical application of the model, allowing users to easily input data and receive predictions on the enthalpy of the combustion of MSW fuel.

Keywords

• Gradient Boosting Regression Tree
• machine learning
• municipal solid waste
• higher heating value