Assembled buildings are now broadly used in the construction industry because of their rapid construction speed and low environmental pollution. In this study, four machine learning model-SVM, RF, XGBoost and CNN-were used to predict the strength of the steam-cured concrete. The results demonstrated that the XGBoost model can better predict the strength, and the R2 and MSE can reach 0.954 and 18.03, respectively. Meanwhile, the importance of steam curing process parameters and the effects on steam-cured concrete properties were investigated via utilizing SHAP analysis and partial dependence analysis. The importance of steam curing parameters in order of priority is as follows: steam curing time, heating rate, steam temperature, pre-curing time, cooling rate, and pre-curing temperature. In the early period (0–7 d), a heating rate of 25 °C/h and a steaming temperature greater than 60 °C had a positive effect on the strength. In the later period (more than 28 d), the pre-curing time had a negative impact on the strength when it was longer than 2 h. Additionally, this study employs the NSGA-III to perform multi-objective optimization of strength, carbon emissions, and cost for steam-cured concrete. At 3 d, the optimal steam curing process parameters were: pre-curing temperature 30 °C, pre-curing time 3.94 h, steaming temperature 50 °C and steam curing time 4 h. At 28 d, the optimal steam curing process parameters were: pre-curing temperature 30 °C, pre-curing time 1.17 h, steaming temperature 45 °C and steam curing time 4 h. This study proposes an approach for enabling intelligent design and low-carbon sustainable production of steam-cured concrete.
扫码关注我们
求助内容:
应助结果提醒方式:
