Early estimation of the long-term deflection of reinforced concrete beams using surrogate models

•The study established a smart ML model to accurately predict long-term deflection of RC beams, abbreviated as WFR-FBI-LSSVR.•217 experimental data samples of long-term deflection were collected to the WFR-FBI-LSSVR and its comparative approaches.•10-fold cross-validation results showed the WFR-FBI-LSSVR to achieve the greatest accuracy with 15.21% of MAPE value.•WFR-FBI-LSSVR attained prediction-to-actual ratio values (1.01) closest to 1, with a minimal 0.23 deviation.