Edge-cloud collaborative digital twin architecture for bridge structural health monitoring: multi-sensor fusion and anomaly detection algorithm design

As the core nodes of transportation networks, bridges face critical challenges in their Structural Health Monitoring (SHM) that are crucial for ensuring transportation safety and extending their service life. Traditional SHM methods encounter high real-time data processing delays, inefficient fusion of multi-source heterogeneous data, and insufficient synchronization between Digital Twin (DT) models and physical entities. Existing DT architectures often lack a hierarchical edge-cloud collaboration mechanism, with fusion algorithms confined to single-modality data, making it difficult to achieve both low latency and high accuracy in anomaly detection. To address these issues, this study proposes a hierarchical edge-cloud collaborative DT architecture, encompassing the physical layer, edge layer, cloud layer, DT layer, and application layer. The core algorithms include: hierarchical multi-sensor fusion; a collaborative anomaly detection framework combining edge-based attention 1D CNN with cloud-based variational autoencoder plus bidirectional long short-term memory network; and Bayesian model updating for dynamic synchronization between virtual models and physical entities. Experimental verification based on a 1:5 scaled steel simply supported beam shows that the architecture achieves a low edge local task delay of 39.7ms and a cloud global task delay of 20.1ms; after strain data fusion, the Root Mean Squared Error (RMSE) is reduced by an average of 62%, and the RMSE for displacement reconstruction is reduced by 68%; the accuracy of anomaly identification reaches over 96%; and Bayesian updating improves the accuracy of strain prediction by 40.1%. This study is the first to construct a hierarchical edge-cloud collaborative DT architecture, filling the technical gap in hierarchical fusion and collaborative computing for bridge SHM. It provides a data-driven intelligent solution for bridge lifecycle health management and is of great significance for promoting the digital transformation of transportation infrastructure.