Evaluation of Concrete Characteristics Using Smart Machine Learning Techniques—A Review

Concrete is one of the most commonly used materials for a wide range of construction across the world. The heterogeneity of concrete results in wide variation in its properties. How different ingredients are mixed, determines the performance of concrete, especially its compressive strength. Hence, rigorous testing of concrete in the laboratories before it is finally selected to be used for a specific type of construction is required. This process of testing requires large quantities of material, time and money. The advent of machine learning and artificial intelligence appears to be promising and significant work has been done in this field for the development of high-performance concrete. Traditional curve fitting models provide the ability to interpolate data, however, this paper evaluates the efficacy of machine learning models and different types of neural networks that have been specifically utilized to predict in terms of various factors, the compressive strength of concrete. Various algorithms like group method of data handling (GMDH) type neural networks, adaptive neuro-fuzzy inference system (ANFIS), hybrid modified firefly algorithm with artificial neural networks (MFA-ANN) and other hybrid ANN algorithms have been used to perform a comparatively analyzed review.