Prediction of self-healing properties of concrete modified with bacteria and fibers using Machine Learning

Abstract

Abstract Self-healing concrete has been studied as an alternative material to overcome problems such as cracking and low durability of conventional concrete. However, laboratory experiments can be costly and time-consuming. Hence, Machine Learning algorithms can assist the development of better formulations for self-healing concrete. In this work, Machine Learning (ML) models were developed using Multiple Linear Regression (MLR), Support Vector Machine (SVM) and Random Forest Regressor (RF) to predict and analyze the repairing rate of the cracked area of self-healing concretes containing bacteria and fibers in their formulations. The results show that the Radial-Basis (RBF) SVM (R2 = 0.927, MAE = 0.053 and RMSE = 0.004) and RFG (R2 = 0.984, MAE = 0.019, RMSE = 0.000) algorithms performed better in predictions and delivered better-fitted models. Therefore, RF regressor and RBF SVM models can be applied to develop and validate high performance self-healing concrete formulations based on polymeric fibers and bacteria.