Integrating computer vision techniques with finite element phase field damage analysis

Abstract

Realistic and accurate finite element (FE) models are crucial for understanding and predicting the health, performance, and safety of deteriorated structures. Accordingly, this paper presents a novel approach that integrates computer vision techniques and a phase field method to enhance FE damage analyses. Computer vision techniques are employed to analyze the visual inspection or monitoring data and to extract the geometric features of a structure and its damage, while the phase field method provides a robust numerical solution for representing the damage and simulating its progression. The integration of these methods allows for automated and precise updates of damage information in the FE model, improving model accuracy and reducing manual intervention. Case studies on a paper board and a steel cross beam of a bridge demonstrate the applicability and effectiveness of the proposed approach, highlighting its feasibility for monitoring and assessment in real-world engineering applications.