Analysis of concrete properties through machine learning models
Abstract
Machine Learning (ML) models, the most prominent methodologies, are now being employed in practically all fields to address difficult and complex problems with a coding-free solution. ML has recently been used in enormous civil engineering applications including cost analysis during construction phase, workforce management in construction site, monitoring the structural health and building life cycle, construction waste management, analysing the mechanical properties (compressive, axial strength, etc.), shear strength and incorporation of various fibers/polymers/demolition wastes in the concrete. This review paper investigates the applications of ML models particularly XGBoost, ANN, RF and SVM used to predict the values of concrete properties (compressive and shear strength) most precisely. The interpretation of input variables across different models is diminished due to constrained datasets. The emplacement of ML models in workplaces is challenging due to the scarcity of datasets pertaining to structures in natural environments. The knowledge gaps and recommendations to enhance the research were also reviewed in this work.