A slab track is used for the primary railway track structure for shinkansen, and it is composed of a concrete bed, filling layer (cement asphalt mortar), RC track slab, and so on. It has been reported that some aged slab tracks laid after the 1980s have voids between the RC track slab and filling layer due to degradation of the filling layer. In the maintenance of the slab track, it is important to detect such voids at an early stage and repair them appropriately. In this study, we investigated a method for detecting voids by non-destructive testing by impact acoustics. As a result, we found that impact acoustics generated in this testing can be influenced by hammering position and supporting condition as well as the existence of voids. Furthermore, it was considered that supervised machined learning is an effective approach for the detection of voids. However, we do not consider the independence of learning data and test data in this study. It is necessary to be verified separately about the discrimination performance for the test data with clear independence such as the measured data from another slab tracks.
View full abstract