机器学习驱动的绿色混凝土强度预测与BIM可视化研究综述OA
A Review of Machine Learning-Driven Strength Prediction and BIM-Based Visualization for Green Concrete
绿色混凝土组成复杂,其强度受再生骨料取代率、矿物掺合料活性及水胶比等多因素耦合影响,呈现强非线性特征,传统经验模型难以实现精准预测.文章系统论述了机器学习在绿色混凝土强度预测中的研究进展,对比分析了 BP 神经网络、SVR 和 XGBoost 三种典型模型的预测性能与适用条件.结果表明,XGBoost 等集成学习模型在大规模数据集上表现最优.基于此,文章阐述了 SHAP 可解释分析方法在揭示材料参数影响机制方面的应用现状,并探讨了 BIM-AI 融合驱动的性能可视化路径.
Green concrete has a complex composition,and its compressive strength is influenced by the coupled effects of multiple factors,including recycled aggregate replacement ratio,supplementary cementitious materials reactivity,and water-to-binder ratio,exhibiting strong nonlinear characteristics that make accurate prediction difficult using traditional empirical models.This paper systematically reviews the research progress of machine learning in green concrete strength prediction,and compares the predictive performance and applicability of three typical models:Back Propagation neural network(BPNN),Support Vector Regression(SVR),and XGBoost.The results indicate that ensemble learning models such as XGBoost demonstrate optimal performance on large-scale datasets.On this basis,the application status of the SHAP explainability analysis method in revealing the influence mechanisms of material parameters is elaborated,and the BIM-AI fusion-driven pathway for performance visualization is discussed.
张帆;罗森林
阜阳职业技术学院城乡建设学院(乡村振兴学院)阜阳职业技术学院城乡建设学院(乡村振兴学院)
建筑与水利
绿色混凝土机器学习可解释分析BIM可视化
green concretemachine learningexplainability analysisBIM visualization
《城市建筑》 2026 (18)
9-12,4
安徽省教育厅自然科学重点项目"数字化驱动BIM-AI融合:绿色建筑安全评估与混凝土性能预测创新研究"(2025AHGXZK30273)
评论