Machine Learning Applications in Nondestructive Testing of Concrete Structures

Abstract Machine Learning bears great potential for data‐driven solutions in the field of nondestructive testing (NDT) concrete structures. The analysis data collected with NDT methods, such as ultrasonics, impact‐echo and ground penetrating radar, can be complex requires experience. expected benefit applications this context goes beyond increase efficiency obtained by automating process. While traditional approaches are usually solely based on key features according to basic principles, algorithms consider full content reveal hidden correlations. For an organized approach data, tool has been developed. provides a graphical user interface manage label training/test interactively define Deep Neural Network architecture. In particular, Convolutional Networks, proven successful various image recognition tasks, implemented. concepts demonstrated show cases, comprising ultrasonic well applications. relevance targeted preprocessing is addressed, comparing effectiveness time‐, frequency‐ joint‐time‐frequency‐representations.