ARTIFICIAL INTELLIGENCE IN STEEL STRUCTURES: METHODS, APPLICATIONS AND PERSPECTIVES ON JOINT BEHAVIOUR

Abstract

<p>Artificial intelligence (AI) is emerging as a powerful tool in structural engineering, enabling data-driven alternatives to traditional analytical and numerical methods. In steel structures, AI has been applied to design optimisation, structural health monitoring, and reliability assessment; however, its application to modelling nonlinear joint behaviour remains limited. This paper reviews AI methods in steel structures, outlining their purpose, capabilities, and current challenges. Focus is given to steel joints, where accurate prediction of moment&ndash;rotation (<em>M</em>&ndash;<em>&theta;</em>) response is essential. A framework combining Genetic Algorithms (GA) for component characterisation and Physics-Informed Neural Networks (PINNs) for surrogate modelling and joint behaviour prediction is presented. The approach enables fast and physically consistent prediction of full nonlinear joint behaviour, supporting advanced AI structural design.</p>