Dynamic crack detection of masonry structures based on digital twins and deep learning

The complexity of material properties, constitutive relationships, and geometric construction of masonry structures leads to a cumbersome and intricate process for analyzing their seismic performance. Moreover, existing methods cannot effectively identify the dynamic crack evolution of structures or accurately track the closure of cracks that cause structural damage. To improve the efficiency and accuracy of damage assessment for structural seismic performance, this paper proposes a crack evolution detection method for masonry structures based on digital twins. To accurately reflect the nonlinear mechanical behavior of the structure, a dynamically updated digital twin model of masonry structures is established by integrating structural monitoring data with refined finite element modeling. The shaking table test of masonry structures provides a basis for calibrating and validating the digital twin model, and the results show that the model can accurately simulate the response and damage state of the structure under seismic action. By comparing the digital twin model with the shaking table test, the crack evolution of masonry structures under seismic wave action is analyzed. The proposed method can accurately track the crack variation pattern, and the accuracy of dynamic crack tracking and detection in masonry structures can reach 96.2%.