Statistical Assessment of Long-Term Bridge Deterioration Using Field-Measured Structural Response Data

Abstract

This study aimed to assess the impact of environmental conditions on bridge deterioration by leveraging a comprehensive dataset comprising 1340 records and 15 variables. The dataset included numeric variables such as acceleration, temperature, humidity, wind speed, and categorical variables like bridge and sensor IDs, structural condition, and damage class. Using advanced predictive modeling techniques, we integrated key environmental variables to enhance the understanding of their influence on structural integrity.

Our methodology involved analyzing the dataset to identify patterns of bridge deterioration, focusing on the role of environmental factors as critical determinants of structural health. We employed machine learning algorithms to forecast deterioration over a 30-day period, utilizing forecast scores as a primary outcome measure. The results showed a significant correlation between environmental conditions and bridge degradation, with higher average degradation scores in structures exposed to increased temperatures (mean 25.186°C), humidity (mean 59.836%), and wind speeds (mean 7.602 m/s).

Bridges classified under "Severe" damage exhibited notably higher forecast scores, indicating accelerated deterioration compared to those with "Minor" or "No Damage." These findings suggest that integrating environmental data into predictive models can significantly enhance the accuracy of deterioration forecasts, aiding in the prioritization of maintenance activities.

Our findings suggest that predictive models incorporating environmental variables offer a robust framework for infrastructure management. This approach facilitates data-driven decision-making, enabling more effective maintenance scheduling and resource allocation. Ultimately, the integration of environmental monitoring into bridge maintenance strategies is crucial for optimizing infrastructure resilience and longevity in the face of evolving environmental challenges.