Flaking concrete detection with hammering inspection methodology based on machine learning

Recently, efficient infrastructure maintenance methodologies have been required in Japan, because the number of aged infrastructures, such as bridges, roads, and tunnels, are increasing drastically. Actual maintenance works consist of a visual inspection, hammering test and paper based archiving. However, there are technical issues, such as a rapider education for professional inspection, infrastructure evaluation cost improvement, and shortage of skilled engineers for infrastructure maintenance. In this paper, we aimed to propose an inspection methodology based on machine learning for concrete structure maintenance using Geographic Information Systems (GIS). We also focused on a hammering test for flaking concrete detection as GIS attribute data acquisition on site. The hammering inspection methodology can evaluate health a condition of concrete surface with hammering sounds. In this research, we have applied a machine learning methodology with k-nearest neighbor (k-NN) algorithm for concrete hammering inspection works.