We are pleased to share the latest scientific publication by our PhD candidate Anastasios Stamou, co-authored with Taniya Kapoor and Prof. Michalis Fragiadakis, recently published in the journal Engineering Applications of Artificial Intelligence.
The paper, titled “A Unified Causality-Enhanced Separable Physics-Informed Neural Network for Predicting Beam and Plate Dynamics”, presents a novel artificial intelligence framework designed to improve the prediction of structural behaviour in engineering applications.
The research introduces an enhanced Separable Physics-Informed Neural Network (SPINN) framework that combines causal training and auxiliary variable strategies for structural dynamics problems. In addition to predicting structural response, the proposed approach enables reliable estimation of engineering quantities derived from spatial and temporal derivatives, providing a more comprehensive assessment of model performance from an engineering perspective.
The framework was systematically evaluated against existing SPINN and Physics-Informed Neural Network (PINN) approaches across a range of forward and inverse structural mechanics problems. The results demonstrate significant improvements in accuracy, convergence, and robustness, highlighting the potential of advanced AI methods for structural analysis and monitoring applications.
This publication represents another important contribution to the growing role of artificial intelligence in structural engineering and infrastructure monitoring, supporting the development of more reliable and efficient data-driven engineering solutions.
