Since the piers in ports are used for logistics activities, it is difficult to stop such operations to inspect the piers, both for economic reasons and also for avoiding disruption to citizens' daily lives, as large quantities of supplies are loaded/unloaded and transported every day.

In this study, we obtained AE data from the operation of mobile cargo-handling machinery as it traveled regularly from one location to another, and performed analyses for formulating a measurement method for evaluating the degree of damage and developing a way of automatically determining the level of damage by using data obtained from AE activities.

For data measurement, we drove the vehicles around while setting the area of damage or non-damage as a parameter. By using the data obtained from multiple sensors (four or more in this project) to detect the elastic waves that occur during such vehicle operation, we identified the points of excitation where elastic waves originated from the damaged locations. We then categorized the obtained data into several different clusters, and formulated a method for determining the degree of damage. More specifically, we adopted κ-means clustering, which is an unsupervised machine learning method, and used it to conduct the cluster categorization using the feature amounts extracted from those elastic waves. As a result, we found that by conducting the cluster categorization based on the feature amounts of the data obtained from the power spectrum (Figure 1), the cluster categorization could be done based on the degree of damage detected. In particular, as shown in Figure 2, when we assumed the number of clusters was four and applied κ-means clustering, we could achieve cluster categorization of varying degrees of damage that corresponded to the mean AE energy and the number of AE hits.

In the future, we will study measurement methods that can take actual site conditions into account, develop a system to be able to make realtime determination, and also research location information systems.