Building information modeling-based bridge health monitoring for anomaly detection under complex loading conditions using artificial neural networks

This study developed an Industry Foundation Classes (IFC) building information modeling (BIM) based framework for bridge health monitoring using behavioral prediction under complex loading conditions. The proposed framework predicts the behavior of the current bridge state under complex loading conditions then employs an anomaly detection method that compares the measured behavior of the bridge structure with the predicted normal value under the same loading condition. This behavioral prediction is accomplished using an artificial neural network (ANN) model based on structural analysis theory and trained using long-term sensor data. The proposed framework operates in an IFC-BIM environment to facilitate bridge management. The IFC spatial element provides a connection between the sensor and the bridge element and between the anomaly information and the IFC object of the bridge element. The proposed framework is then demonstrated on a field cable-stayed bridge in Korea. The results confirm the prediction accuracy of the proposed ANN model under complex loading conditions and its ability to identify element anomalies for maintenance.