Elaboration and validation of IP-ANN model of fly ash-clayey soils

This study aims to develop an Artificial Neural Network (ANN) model to predict the Plasticity Index (PI) of clay soils stabilized with fly ash. A total of 193 datasets were collected from previous studies, normalized, and used for training, testing, and validating the model using the backpropagation algorithm. The results indicated that the PI-ANN model achieved high accuracy, with a correlation coefficient exceeding 99% and minimal Mean Square Error (MSE). Sensitivity analysis showed consistency with existing literature. Validation using experimental results from three independent studies confirmed the model’s performance, achieving a regression coefficient of 99% and a mean relative error of 0.92%. Additionally, the PI-ANN model outperformed a comparable statistical model, demonstrating its reliability for predicting the PI of high-plasticity clay soils.