Trident: A Deep Learning Framework for High-Resolution Bridge Vibration Monitoring

Bridges are the essential components in lifeline transportation systems, and their safe operation is of great importance. Information on structural damage could assist timely repairs reduce downtime. With latest advancements sensing technology, collecting vibration data from bridges has become more accessible. However, effective processing still a challenge, given high dimensionality massive size data. Existing studies have shown that machine/deep learning techniques can be valuable tools for this task. computational capacities these models challenged presence large sensor arrays. We propose Trident as novel deep framework enables automatic feature extraction by simultaneously temporal three-dimensional (3D) spatial variations 6D input instrumented bridges. equipped with 3 ConvLSTM3D branches to achieve goal. A 3D steel truss bridge subject dynamic traffic loads monitored its vibrations evaluate Trident’s robustness finding damaged elements. dataset 52,800 vehicle passing simulations generated leveraging database 528 passenger vehicles United States, obtained National Highway Traffic Safety Administration. Bayesian optimization utilized tune model’s hyperparameters, achieving test Node Average Geometric Mean Accuracy 86%. This level performance promising complexities output space vibration-based monitoring. concept extended other monitoring tasks different time series labeling strategies.