A digital twin framework for turbine blade crack propagation prediction

Abstract

Digital Twin offers a novel methodology for structural health monitoring (SHM) across various fields. This paper proposes an integrated digital twin framework that combines monitoring and simulation for SHM and life prediction of critical structural components. The framework incorporates the Mask R-CNN network to extract damage-related features from structural response field images and employs the dynamic Bayesian network (DBN) coupled with parametric modeling for real-time model updating. A custom-developed visualization platform enables real-time representation of digital twin model. As a case study, the proposed framework is applied to turbine blades crack propagation, involving physical experiments, automated crack detection, digital model updating, and crack propagation life prediction. The results show that the framework achieves accurate crack identification, with a maximum inversion error of 10.5 %, and reliable life prediction, with a final error of 7.15 %. This study offers an efficient and practical approach for SHM and life prediction, offering significant potential for intelligent structural monitoring and predictive maintenance in engineering applications.