This study presents an artificial neural network (ANN) framework for predicting fretting-induced damage in cylinder-on-flat geometries representative of submarine power cable conductors. The ANN is trained using data generated by EASY-FRET (Engineering Analysis SYstem for FRETting), a theoretical fretting analysis tool based on a two-dimensional cylinder-on-flat configuration that accounts for geometry, loading and material properties. The framework achieves over 90% prediction accuracy while reducing computational time from days, using finite element (FE) models, to minutes, with EASY-FRET, and, ultimately, to seconds, with the ANN. This approach provides a foundation for future digital twin development and represents a significant improvement over traditional finite element or experimental methods, which require significantly longer times (e.g. years) to generate comparable datasets.
Poon et al. (Mon,) studied this question.