Developing Soft-Computing Models for Simulating the Maximum Moment of Circular Reinforced Concrete Columns

Abstract

There has been a significant rise in research using soft-computing techniques to predict critical structural engineering parameters. A variety of models have been designed and implemented to predict crucial elements such as the load-bearing capacity and the mode of failure in reinforced concrete columns. These advancements have made significant contributions to the field of structural engineering, aiding in more accurate and reliable design processes. Despite this progress, a noticeable gap remains in literature. There's a notable lack of comprehensive studies that evaluate and compare the capabilities of various machine learning models in predicting the maximum moment capacity of circular reinforced concrete columns. The present study addresses a gap in the literature by examining and comparing the capabilities of various machine learning models in predicting the ultimate moment capacity of spiral reinforced concrete columns. The main models explored include AdaBoost, Gradient Boosting, and Extreme Gradient Boosting. The R2 value for Histogram-Based Gradient Boosting, Random Forest, and Extremely Randomized Trees models demonstrated high accuracy for testing data at 0.95, 0.96, and 0.95, respectively, indicating their robust performance. Furthermore, the Mean Absolute Error of Gradient Boosting and Extremely Randomized Trees on testing data was the lowest at 36.81 and 35.88 respectively, indicating their precision. This comparative analysis presents a benchmark for understanding the strengths and limitations of each method. These machine learning models have shown the potential to significantly outperform empirical formulations currently used in practice, offering a pathway to more reliable predictions of the ultimate moment capacity of spiral RC columns.

Keywords

• circular reinforced concrete columns
• moment capacity
• machine learning models
• soft-computing techniques