Ensemble Learning Based Approach for Corrosion in Structural Concrete

This research focuses on steel corrosion in concrete structures, a major concern in the construction industry. The aim of the project is to use state-of-the-art machine learning techniques, particularly ensemble approaches, to increase the reliability and precision of corrosion prediction models for reinforced concrete structures. Temperatures, relative humidity, anode lengths, and concrete age are examples of critical input characteristics that were ascertained by meticulous data collection and testing. To evaluate the efficacy of the various ensemble learning techniques—Boosted Trees, Bagged Trees, and Optimizable Ensembles—performance metrics such as RMSE, R-squared, MSE, MAE, prediction speed, and training time were employed. Bagged trees outperform Boosted trees in terms of R-squared (0.96 vs. 0.85), lower RMSE (17.393 vs. 33.784), MSE (302.53 vs. 1141.4), and MAE (12.642 vs. 29.594). They also have equal prediction speeds and training timeframes. Bagged Trees perform better than other ensemble approaches in terms of prediction accuracy, training efficiency, and general usability, whereas optimization ensemble methods offer competitive results with potential benefits in feature selection and dimensionality reduction. Time-sensitive applications can take advantage of the superior performance of Boosted Trees and Bagged Trees, which balance forecast accuracy with speed.