Artificial Intelligence-Based Prediction of Vibration Frequencies in Cold-Formed Steel Plates

Cold-formed steel plates are a highly versatile component in modern structural systems due to their lightweight nature, high strength, and structural efficiency. These characteristics make them particularly suitable for sustainable and high-performance construction. However, it is essential to understand their vibrational behaviour in order to prevent resonance phenomena and ensure the dynamic stability and safety of structures. Traditional numerical approaches, such as the Finite Strip Method (FSM), are commonly used to determine natural vibration frequencies with reliable accuracy. However, these methods can be computationally demanding and time-consuming, particularly when conducting extensive parametric studies with multiple geometric configurations.

In order to address this limitation, this study proposes the use of artificial intelligence techniques for the prediction of vibration frequencies of cold-formed steel plates. A complete database was created using CUFSM software. Key geometric parameters such as plate length (a), width (b) and thickness (t), along with the geometric ratios (a/b) and (b/t), were varied. The generated dataset was then utilised to train and evaluate multiple machine learning models, encompassing linear regression, support vector regression (SVR), and the Random Forest algorithm.

The results obtained demonstrate the capacity of machine learning models to predict vibration frequencies with a high degree of accuracy, while significantly reducing computational time in comparison to conventional numerical methods. Following rigorous testing, the Random Forest algorithm has demonstrated its particular strength in performance when compared with other models. This approach is a reliable solution for structural engineering applications that require rapid prediction and optimisation.