Predicting fatigue life of shear connectors in steel‐concrete composite bridges using artificial intelligence techniques.

Fatigue limit states often govern the design of shear connectors in steel‐concrete composite bridges. AASHTO LRFD bridge design specifications provides a linear equation in a semi‐logarithmic S‐N curve for predicting the fatigue life of shear connectors. However, this equation can be too conservative in some cases, as supported by the available experimental data. In this paper, artificial intelligence (AI) was incorporated into the prediction of the fatigue life of shear connectors. Six different machine learning (ML) algorithms were considered for this purpose. The predictions of ML algorithms were compared both with the available experimental data and the equation provided by AASHTO. The results showed that the fatigue life predicted by ML methods is more accurate than that predicted by the current equation of AASHTO. The results of this study showed that AI can be a proper alternative to the existing methods for predicting the fatigue life of shear connectors. Highlights: AI was used for the first time to evaluate fatigue life of shear connectors in bridges.AI predicted fatigue life of shear connectors more accurately than the current methods.GPR and DT algorithms were the best algorithms for assessing fatigue life of connectors.The current equations in AASHTO for fatigue may need to be redeveloped based on AI. [ABSTRACT FROM AUTHOR]