Prediction of Interface Friction Angle Between Landfill Liner and Soil Using Machine Learning

This study employs machine learning (ML) techniques and artificial neural networks (ANN) to predict the interface friction angle between the landfill liner and the soil. The interface behavior is majorly affected by the thickness of landfill liner (t), mass of landfill liner (m), tensile strength of landfill liner (T), cohesion of soil (cu), angle of shearing resistance of soil (Φ), shear strength (τ), and normal stress (σ). As the stability of landfill liner varies significantly from that of the soil due to the non-homogeneity and anisotropic character of the soil, it is critical to comprehend the interface behavior between the landfill liner and the soil. However, no prior research employing machine learning techniques to analyze the interface behavior between landfill liners and soil has been reported; a study using machine learning algorithms and artificial neural networks is carried out on 66 datasets to probe the interface behavior with the help of an ANACONDA navigator. Further, to understand the impact of input variables on the output variable, Pearson’s correlation coefficients were determined. Mean absolute error (MEA) is considered as a loss function, and the best model was chosen based on the r²-value. Random forest regressor (RFR) model is determined to be the best model among the available models with an r²-score of 0.99 and a minimum mean absolute error of 0.46.