Browsing by Author "Sidvilasini, S."

Now showing 1 - 3 of 3

A Hybrid Machine Learning Approach for Predicting Joint Shear Capacity in Beam-Column Connections
(Springer Science and Business Media Deutschland GmbH, 2025) Sidvilasini, S.; Palanisamy, T.
Accurately predicting the shear strength of beam-column connections is crucial for maintaining the structural integrity and stability of buildings, especially in seismic conditions. This study aims to address this challenge by developing and evaluating multiple machine learning regression models for estimating joint shear capacity. A dataset consisting of 445 beam-column connections with 17 key influencing variables was compiled and used to train seven distinct regression models. Among them, the four best-performing models—Quadratic Support Vector Machine (QSVM), Rational Quadratic Gaussian Process Regression (RQGPR), Kernel Ridge Regression (KRR), and Ensemble Boosting (EB)—were selected based on their predictive accuracy. To further enhance performance, these models were combined into a hybrid ensemble model, capitalizing on their complementary strengths to improve shear strength estimation. The hybrid model exhibited superior predictive performance, achieving a test RMSE of 0.0246 and an R2 value of 0.9605, significantly surpassing the accuracy of the best standalone model (RQGPR). This reinforces the advantage of ensemble learning in minimizing error and enhancing generalization. The findings of this research highlight the growing role of machine learning in structural engineering, particularly in advancing shear strength prediction methodologies. By demonstrating that a hybrid model can outperform traditional single-model approaches, this study provides valuable insights for developing safer, more resilient structures and optimizing modern engineering practices with artificial intelligence. © The Author(s), under exclusive licence to Shiraz University 2025.
Improving Structural Safety with Machine Learning: Shear Strength Prediction in Interior Beam-Column Joints
(Institute of Electrical and Electronics Engineers Inc., 2024) Sidvilasini, S.; Palanisamy, T.
Determining the shear capacity of joints between columns and beams within a structure is crucial to guarantee its safety and stability. It directly impacts buildings' structural integrity, cost-effectiveness, and resilience, making it a critical aspect of structural engineering and construction. Estimating shear properties in beam-column joints is done via machine learning due to its ability to capture complex relationships, adapt to diverse data, and automatically identify relevant features, potentially offering improved accuracy and insights compared to traditional methods. This paper includes creating a machine-learning regression model for predicting joint shear strength in interior beam-column joints. It involves the analysis of a comprehensive dataset comprising 445 data points with 17 variables sourced from 100 research papers. The primary objective is to craft a machine-learning regression model capable of accurately forecasting joint shear strength. To achieve this goal, a multitude of methodologies have been explored, including the application of 2 machine learning regression techniques and two codes of practice (Step-wise Linear Regression, Medium neural networks, and EN 1998-1:2004, NZS 3101:1-2006). Of the two methods, step-wise linear regression gave the best results in predicting the shear capacity of interior column-beam connections. Â© 2024 IEEE.
Predicting joint shear in beam–column connections using convolutional neural networks
(Springer Science and Business Media B.V., 2025) Sidvilasini, S.; Palanisamy, T.
Predicting joint shear at beam-column junctions (BCJ) is essential in structural engineering to ensure the safety and reliability of systems. Current methodologies using empirical calculations may rely on simplistic assumptions and insufficiently account for the many geometric factors and material properties that influence shear in BCJ. This research introduces a novel approach using Convolutional neural networks (CNNs) to predict joint shear. The collection comprises 515 joints, categorized into 210 exterior joints and 305 interior joints, characterized by 14 fundamental factors delineating their form and material properties. The predictive performance of the CNN model has been evaluated using known engineering codes, including ACI 318-19, NZS 3101:1-2006, IS 13920:2016, and several other data-driven models in the domain. Furthermore, it has been contrasted with an ensemble regression method. The study includes a thorough sensitivity analysis using a gradient-based method to determine the relative importance of input factors in predicting shear stress. The findings demonstrate the effectiveness of CNN in identifying complex relationships among joint parameters, thereby enabling precise predictions of joint shear. This method offers a promising alternative to traditional empirical formulas and enhances the understanding of structural behavior in BCJ. This study integrates contemporary machine learning algorithms with structural engineering concepts to optimize design processes and augment the safety and reliability of built environments. © The Author(s), under exclusive licence to Springer Nature Switzerland AG 2024.