AI-supported condition analysis of existing timber structures with deep learning
The content on this page was translated automatically.
The engineering assessment of existing timber structures requires a precise condition assessment as a basis for load-bearing capacity analyses and repair concepts. However, visual inspection of long-span glulam beams often involves a great deal of effort. Building on an existing AI methodology for image-based object recognition of nail connections, the method is being extended to the automated detection and evaluation of cracks. Using deep learning methods, cracks are identified, geometrically analyzed and integrated into digital 3D models. This enables a data-based condition assessment and supports the planning of repair measures. In addition, methods are being developed to reliably identify intact partial cross-sections from damaged beams and assess their potential for reuse in new timber constructions. The project thus contributes to conserving resources and strengthening the circular economy in timber construction.