A Comparative Analysis of Machine Learning Algorithms for Predicting Fundamental Periods in Reinforced Concrete Frame Buildings

Abstract

Determining the fundamental period is a critical aspect of the analysis and design of structures. Existing literature formulas for evaluating this parameter exhibit a wide range of variations. To tackle this problem, various algorithms and techniques are used to learn patterns and relationships from data, enabling to make more precise predictions. In the present study, a dataset consisting of 162 RC frame-building models was analyzed using ETABS 2016. The fundamental period, a critical output parameter, is predicted using four machine-learning models: Support Vector Machine (SVM), Random Forest (RF), Artificial Neural Network (ANN), and Gradient Boosting Regression (GBR). The performance and accuracy of these models were compared to identify the best-performing model for predicting the fundamental period in structural analysis. The efficiency and precision of the machine learning algorithms were assessed based on the R² and root mean square error (RMSE) values. Among all the models, the GBR exhibited the best performance with an R² (Coefficient of determination) score of 0.9995 and an RMSE of 0.017. These results indicate that the GBR model achieved a high level of accuracy and a low level of prediction error in estimating the fundamental period of the RC frame-building models.