Comparison of Random Survival Forest with Accelerated Failure Time-Weibull Model for Bridge Deck Deterioration

Bridge deck deterioration modeling is critical to infrastructure management. Deterioration traditionally done using deterministic models, stochastic and recently basic machine learning methods. The advanced learning-based survival such as random forest, have not been adapted for use in This paper introduces forest models bridge compare their performance with a commonly used traditional model, that is, the Weibull distribution-based accelerated failure time (AFT-Weibull) model. To better adapt model modeling, selection of dependent variables discussed between two variables: time-in-rating, cumulative truck traffic. Inspection data from about 22,000 state-owned decks Pennsylvania are validate test models. results suggest traffic more suitable be selected variable when analyzing reliability deck. Further, outperformed AFT-Weibull predictive accuracy.