摘要
混凝土碳化深度是影响结构耐久性的重要因素之一,精准的预测混凝土碳化深度,对工程施工具有一定的现实意义。为探究混凝土碳化深度与其影响因素之间的关系,使用鲸鱼算法(WOA)对BP神经网络的权值、阈值进行优化处理,并以此建立了6输入变量、9神经个数和1输出变量的拓扑结构。以相对湿度、粉煤灰和水泥含量、水胶比、CO 2浓度及龄期为输入变量,混凝土碳化深度作为输出变量,对BP神经网络和WOA-BP神经网络进行预测结果对比分析。结果表明,神经网络模型预测碳化深度是可行的。
Concrete carbonation depth is one of the important factors affecting structural durability,and accurate prediction of concrete carbonation depth has certain practical significance for engineering construction.To investigate the relationship between concrete carbonation depth and its influencing factors,the weights and thresholds of the BP neural network were optimized using the Whale Algorithm(WOA),and a topology of 6 input variables,9 neural numbers and 1 output variable was established.Meanwhile,the prediction results of BP neural network and WOA-BP neural network were compared and analyzed with relative humidity,fly ash and cement content,water-cement ratio,CO 2 concentration and age as input variables,and concrete carbonation depth as output variable.The results show that the neural network model is feasible for predicting the carbonation depth.
作者
刘勋飞
桂泽人
甘坚宇
梁帅锋
Liu Xunfei;Gui Zeren;Gan Jianyu;Liang Shuaifeng(Construction Engineering Quality and Safety Supervision Station of Shenzhen Pingshan District,Shenzhen Guangdong 518000,China;China Construction Science and Industry Corporation Ltd.,Shenzhen Guangdong 518000,China)
出处
《山西建筑》
2023年第12期125-129,共5页
Shanxi Architecture
基金
中建科工技术研发项目(No.ZJKG/HN/2021-026/FBHT/JSZX-01)。
