Artificial Neural Network Method for Predicting Compressive Strength of Normal Concrete

Abstract

Lombok Island is an archipelago that has a source Natural resources such as sand and gravel are abundant. This material is one of the components of concrete. Concrete is a frequently used material in Indonesia. Compressive strength testing of concrete typically requires a large number of samples and a considerable amount of time. To expedite and simplify this process, researchers employ computer-based intelligence techniques, namely the Artificial Neural Network (ANN) method. This research involved a series of laboratory tests for normal concrete's compressive strength. The obtained data was then processed using MATLAB with the ANN modeling method for training. The research results indicated a Mean Absolute Percentage Error (MAPE) of 0.02% during the training process and 1.54% during testing. This demonstrates that the developed ANN modeling exhibits a high level of accuracy with low error. Therefore, the empirical formula obtained can be used for predicting the compressive strength of normal concrete with a good degree of precision.