ArtBridgeNet: artificial neural network for seismic resilience assessment of highway bridge network

The resilience of highway bridge networks is critical to maintaining quality of community life. Bridge networks must be sufficiently resilient to withstand various hazards, mitigate the impacts of catastrophic events, and rapidly recover to pre-event conditions. Bridges ensure continuity of daily commuting and enable the timely delivery of emergency aid and resources to affected communities, thereby accelerating infrastructure restoration. Moreover, bridges should not be considered as independent components, as their performance plays a vital role in preventing or minimizing community-level disruptions and outages. This study evaluates the seismic resilience of a regional bridge network by emphasizing the interconnectedness and interdependence among individual bridges and emergency facilities. An Artificial Neural Network (ANN)–based resilience assessment framework is developed and applied to a bridge network in Vancouver, British Columbia, consisting of 11 bridges across seven major routes. Using a network-based reliability model and simulated earthquake events with varying magnitudes and epicenter locations, resilience and reliability indices are calculated and used to train the ANN. The trained model efficiently predicts resilience and reliability metrics for different seismic scenarios, demonstrating high accuracy with significantly reduced computational cost and time.