In this study, we harnessed the potential of ensemble machine learning models and artificial intelligence (AI) techniques to address the limitations of conventional scour depth estimation methods. Our primary objective is to investigate how these advanced techniques can improve the accuracy of scour depth predictions, enabling more effective bridge engineering and maintenance practices. To achieve this objective, we employed ensemble machine learning (ML) approach comprising CatBoost, random forest, histogram gradient boosting, and extreme gradient boosting algorithms. These models were rigorously trained and tested on field data to predict bridge scour depth accurately. Furthermore, the performance of the ensemble machine learning models was benchmarked against that of an Artificial Neural Network (ANN) model. The key findings of this study show the superior performance of ensemble ML models over the ANN model in predicting bridge scour depth. Among the ensemble models, the CatBoost algorithm had the best performance with R-square value of 0.85 and a minimal root mean square error (RMSE) value of 0.47.
|