Artificial intelligence for the prediction of the physical and mechanical properties of a compressed earth reinforced by fibers

Abstract

The use of natural fibers as a reinforcing product in the production of compressed earth blocks can be considered as an effective means for the environment and savings. This study presents a prediction a model- and simulation-based approach using artificial neural networks (ANN) to predict tensile strength and compressive strength. of earth-friendly concrete containing different types of natural fibers. A group of data with eight influencing characteristics; cement, fiber, sand, fiber length, fiber tensile strength, clay, silt, age used for model formation and validation were collected from the literature. The output was compressive strength and tensile strength. The combination of root mean square propagation and stochastic propagation gradient descent with the momentum method is used to train the ANN. Using various validation criteria such as coefficient of determination (R), root mean squared error (RMSE) and mean absolute error (MAE), the ANN model was validated and compared to two machine learning (ML) Random Forest (RF) techniques and Multilayer Perceptron (MLP). A sensitivity analysis was also performed to validate the robustness and stability of these models. The experimental results showed that the ANN model performed better than other models and, therefore, it can be used as a suitable approach to predict the compressive strength of environmentally friendly earth concrete.