Physics-Assisted Machine Learning for Predicting Splitting Tensile Strength of Recycled Aggregate Concrete

Machine learning algorithms, aided by physics, are being used to further predict the splitting tensile strength of recycled aggregate concrete. Researchers have compiled a database of 257 data points, derived from physical experiments and existing mechanical models. With a total of ten numerical input variables and one non-numerical variable, various machine learning approaches have been used to accurately predict the output variables, specifically the splitting tensile strength of recycled aggregate concrete. After careful examination and training, the RF and Adaboost models demonstrate the best performance. However, it is recommended that ensemble models be used to improve the accuracy and generalizability of the predictions. Advances in machine learning algorithms capable of predicting the tensile strength of recycled aggregate concrete can lead to improved understanding, and the implementation of corresponding products, thereby reducing the waste of natural resources.