Prediction of Mechanical Characteristics of Pebbles Sand Concrete Using Artificial Neural Networks (ANN)

Abstract

An innovative and successful method for producing concrete mixtures that are ideal for their intended usage is artificial neural networks. Several test batches of concrete were made in the lab for this study to assess the mechanical characteristics of pebbles sand concrete, and the outcomes of the trials were documented for predict of mechanical properties of pebbles sand concrete, Data was gathered from the accessible literature and entered into a database. 146 specimens' worth of data were utilised to predict the compressive strength. The split tensile strength was predicted using data from 152 specimens. Since there were few data available, 92 specimen data were utilised to predict the flexural strength. Any type of concrete strength can be precisely predicted utilising controlled software approaches. In this study, MATLAB R2015b were used to predict the mechanical properties of pebbles sand concrete. In ANN, samples are used for testing, validation, and training in proportions of 70%, 15%, and 40%, respectively. All mechanical characteristics of pebble sand concrete were accurately predicted by the generated multi-linear regression models, with correlation coefficients (R² values) which is closer to 1. The scatter plot graphs indicate a strong correlation between the experimental and predicted outcomes.