Internal Damage Detection for Rubber Bearings Using Machine Learning Methods

Abstract

Rubber bearings are now commonly used to prevent seismic damage to bridges, but they can be destroyed in huge earthquakes. Even if it is not destroyed, it may suffer internal damage, however, it is difficult to detect internal damage because it is covered by coated rubber. This study verified the propagation of elastic waves in a laminated rubber bearing by numerical analysis, and used the data to detect damage using a machine learning anomaly detection technique. The results showed that the damage inside the rubber bearing can be detected using the One-Class Support Vector Machine anomaly detection method.