Prediction and validation of constituent materials for concrete manufacturing using artificial neural network

摘要

The development of high-strength concrete is based on the mix proportion determined through mix design. After conductingmultiple trials on the mix proportions, a typical concrete mix with the requisite strength can be achieved. As a result, the proceduretakes far too long to complete a large number of mix design trials. Therefore, the requirement of advanced technologyto save time, manpower, and material is required. This work is mainly focused on the creation of a MATLAB-based artificialneural network (ANN) model for predicting concrete’s compressive strength, determining the projected values of concrete’smechanical characteristics, and conducting a correlation between the results of the experiment and the predicted values. Atotal of 1030 pieces of mixed proportional data were collected from various researchers to train the neural network. And tovalidate the trained data, a total of five mix proportions were prepared as per the Indian standard code for mix design. Fromthe results, there is a good correlation between the trained and experimental data. Furthermore, the error values are foundto be minimal, and the test and experimental data are well correlated (R2 = 0.95).