Prediction the Settlement of the Lightweight Road Embankments Using ANN

Abstract

The effectiveness of the Artificial Neural Network in modelling the link between the settlement of the embankment and embankment parameters is demonstrated in this study. Data were collected from previous research and the Plaxis 3D program. Eleven factors affecting the settlement of embankments were included in this study. These factors are (Side slope & Height(h) of the embankment), (Axial load, and No. of point axial load), in addition to (γ_unsat, γ_sat, Initial void ratio (e), Friction angle (Ø), Cohesion (c), Poisson ratio (ⱱ), and modulus of elasticity (E) which are the properties of the soil under embankment). In the current study, a back-propagation neural network approach was used. The relative importance analysis showed that the soil modulus of elasticity (E) and height (h) of the embankment was the most influential parameters than other inputs. For each training, validation, and testing step, the correlation coefficient R-values of the settlement embankment dataset of the ANN model were obtained to be approximately 0.984. The results revealed the validity of using Artificial Neural Networks to generate settlement embankment soil values.