2026 Volume 5 Issue 1 Pages 239-245
This paper describes a method for inspecting continuous fiber sheet repair sections on actual structures using the “AI-aided Hammering Test System,” developed to establish a percussion inspection technique capable of accurately detecting damage independently of the inspector's skill. This system uses machine learning to analyze differences in percussion sounds produced by an inspection hammer, automatically detecting abnormal areas in the structure and their severity. Furthermore, it conveniently acquires the hitting positions using a laser-based range sensor and automatically generates an anomaly map by integrating these data with the analyzed results. This reduces the workload, including the creation of drawings after the sound inspection. The following sections describe the AI-aided Hammering Test System's configuration, functions, and learning algorithm, and report the results of sound tests conducted on the underside of an actual bridge deck slab.
