Sulfate Freeze–Thaw Damage Evolution and Life Prediction of Recycled Concrete Based on Entropy Weight Method and Grey Theory

The effects of recycled coarse aggregate (RCA) replacement rate, freeze–thaw environment, and cycle number on the durability of recycled coarse aggregate self-compacting concrete (RCASCC) were experimentally investigated. Compressive strength, splitting tensile strength, uniaxial compressive strength loss, and ultrasonic wave velocity loss were analyzed. Results show that within 0–50 freeze–thaw cycles, strength loss of R0, R50, and R100 exhibited no significant differences across four environments, whereas clear stratification appeared during 50–125 cycles. As cycles increased, the splitting tensile strength loss curve gradually flattened, while ultrasonic wave velocity loss reached a minimum between 50 and 75 cycles. Among all environments, 5 wt% MgSO4 solution caused the most severe deterioration. Based on entropy weight theory, a durability evaluation index was established, showing a decreasing trend with increasing cycles and a stratified distribution among groups. The GM(1, 1), Verhulst, and parabolic regression models demonstrated high prediction accuracy, while GM(1, N) performed poorly. Using Inner Mongolia as a case study, service life under combined freeze–thaw and sulfate attack was predicted, with the Verhulst model providing more conservative estimates. Furthermore, a BP neural network optimized by GM(1, 1) and Verhulst models was developed. Compared with GM(1, 1), the Verhulst-optimized BP model showed superior prediction accuracy and generalization ability.