Structural Failure Probability Classification in Ageing Bridge Networks: A Multimodal Logistic Regression Framework Integrating Acoustic Emission Signals, Strain Gauge Readings, and Visual Inspection

The article presents an R-based workflow for classifying structural failure probability in ageing bridge networks using multimodal logistic regression. It combines bridge inventory data, traffic exposure, environmental conditions, visual inspection scores, acoustic emission signals, strain gauge readings and engineered interaction variables.

Using simulated but realistic bridge data, the article shows how ageing, heavy traffic, corrosion, scour, cracking, acoustic activity, abnormal strain response, fatigue damage, maintenance backlog and low load rating can increase structural risk. The workflow includes EDA, spatial maps, logistic regression modelling, model evaluation, coefficient interpretation, calibration, risk deciles, high-risk flagging and visual-sensor disagreement analysis.

Overall, the article demonstrates that bridge failure risk is multidimensional. Logistic regression provides a transparent and interpretable baseline for prioritising bridge inspections and maintenance, but real-world use would require engineering validation, calibrated sensors and expert review.