摘要
Abstract
To predict the carbonation depth of concrete under continuous high-humidity environments,nine groups of concrete specimens with different mix proportions are prepared,and carbonation tests are conducted to systematically investigate the effects of carbonation age,water-binder ratio,fly ash content,slag content,and relative humidity on carbonation depth.Based on experimental data,BP and GA-BP neural network models are established,and their predictive performances are comparatively analyzed.Results demonstrate that the GA-BP model exhibits enhanced prediction accuracy,with its R2 increasing by 0.91%compared to the BP model,while MAE,MSE,and RMSE values all decrease;Carbonation age and water-binder ratio are dominant influencing factors,with relative humidity and mineral admixtures also showing significant effects.This study provides theoretical support and practical guidance for optimizing concrete mix designs in engineering applications under high-humidity conditions.关键词
混凝土碳化深度/GA-BP神经网络/高湿度环境/预测模型/影响因素分析/主成分分析法Key words
concrete carbonation depth/GA-BP neural network/high-humidity environment/prediction model/influencing factor analysis/principal component analysis分类
建筑与水利
