Predicting fatigue damage of highway suspension bridge hangers using weigh-in-motion data and machine learning

Continuous and real-time tension force monitoring is a key point in fatigue damage evaluation for bridge suspenders or hangers. Usually, effective sensors are not equipped in suspenders or hangers of in-service bridges to obtain tension force responses. Bridge-site-specified traffic loading information collected by Weigh-in-motion (WIM) system offers an opportunity to address this issue. The daily fatigue damage of hangers can be estimated by combination of the traffic loading data with finite element analysis. Support vector machine (SVM) is adopted to establish the regression models between daily fatigue damage and collected traffic loading parameters. Consequently, the future fatigue damage of cables or hangers can be predicted by feeding the subsequent WIM data into the regression models. This proposed method is validated in the fatigue life prediction of hangers on a suspension bridge. The SVM model configuration and generalisation ability are investigated in this study. This study presents a novel way to estimate the fatigue damage of the hanger without direct stress sensing equipment and provides new thoughts in interpreting the monitoring data to provide useful information for engineering decision makers.

Automated summary

This study proposes a novel method to estimate the fatigue damage of bridge hangers using support vector machine regression models based on traffic loading data and finite element analysis, eliminating the need for direct stress sensing equipment. The method is validated in fatigue life prediction of hangers on a suspension bridge.