PSC Damage Detection Approach for Tendon of Prestressed Concrete PSC Bridge Based on Auto-Encoder

摘要

The present invention relates to a method for estimating tendon damage of a PSC bridge using an autoencoder, and the present invention relates to acceleration, displacement, strain, etc. representing small behavior of the PSC bridge that can be obtained through the SHM system in order to understand the danger of the bridge in real time. A first step of collecting bridge measurement data for a set period so that the damage of the tendon can be estimated using the bridge measurement data of A second step of generating a tendon damage estimation model capable of restoring only the'non-damaged state' data by performing auto-encoder learning by considering the bridge measurement data as'non-damaged state' data; PSC using an autoencoder comprising; a third step of determining whether the bridge tendon is damaged by determining whether it is'non-damaged state' data through a tendon damage estimation model when the bridge measurement data measured in real time is input from the bridge measurement unit. Provides a method for estimating the tendon damage of a bridge.