ABSTRACT
Prestressed concrete box-girder bridge is the main structural type of the small and medium span bridges on the China highway. Cracking is the main problem for concrete bridges. Benefit from structural health monitoring system, the monitoring data makes it possible to quickly indicate the cracking state of the bridge. Although the vehicle-induced strain responses of each bridge component have randomness and variability, they also obey some certain regularity. Based on the live-load strain data from bridge monitoring system, this study develops a comprehensive method of state evaluation and cracking early-warning for the prestressed concrete box-girder bridge. The feature of vehicle-induced strain is extracted using the deep learning and classification of long short-term memory network. The vehicle-induced strain features are clustered via Gaussian mixture model, and the early-warning of bridge cracking is conducted according to the reliability of heavy vehicle clustering data. This method can be used as an indicator for the bridge inspection, truck-weight-limit and reinforcement work.
