Prediction of the Shear Strength of Glass Fiber-Reinforced Clay Soil by Adaptive Neuro-Fuzzy Inference System (ANFIS)

Abstract

In recent years, the use of artificial intelligence algorithms in geotechnical engineering has increased, and successful results have been obtained in geotechnical engineering using artificial intelligence algorithms. The objective of this study is to estimate the shear strength of glass fiber reinforced clay soil using ANFIS. For this purpose, specimens with different water contents (13%, 15% and 17%) and different glass fiber addition ratios (0%, 1%, 1.5% and 2%) were prepared. The ANFIS models were created using the shear strength (τ) data obtained by direct shear tests on the prepared specimens. To create the best fitting ANFIS model in the current study, 75%, 77%, 80%, and 83% of the data for training and 25%, 23%, 20%, and 17% of the data for testing were used, respectively. However, to estimate the shear strength in each ANFIS model, the normal stress (σ), glass fiber content (Fc), and water content (ω) are considered as input parameters. Statistical parameters such as root mean square error (RMSE), regression coefficient (R2), root square error (RSE), and mean absolute error (MAE) were also calculated to determine the success rates of the ANFIS models. Examination of the statistical parameters revealed that the data used 80% for training and 20% for testing provided the best results in estimating the shear strength of the ANFIS model.