摘要
混凝土耐久性受到许多不确定性因素的影响。基于RBF神经网络结构具有自适应确定性,输出值与初始权值无关的优点,根据影响施工期混凝土材料特性耐久性主要因素,建立了相应的混凝土裂缝深度预测模型。根据已有的试验数据,利用MATLAB数学软件实现了对该模型的训练。利用训练好的模型进行混凝土裂缝深度预测。预测结果表明,预测值与实测值误差相对较小,在允许范围内,所以RBF神经网络预测根据施工期混凝土材料特性能够实现碳化混凝土裂缝深度在实际工程中的预测。
Concrete durability is affected by many uncertainty factors. RBF neural network has the advantage of adaptive certainty and the output value has nothing to do with the initial weights. According to the main factors affecting concrete durability, the prediction model for concrete durability depth is established based on the advantages of RBF neural network. Combining with the MATLAB mathematical software, it is used to test mathematical model by the experimental data. The network is used for the prediction for concrete carbonation. The prediction results show that the forecast results conform to the test results very well. Thus, it can be considered a reasonable method.
出处
《施工技术》
CAS
北大核心
2014年第3期48-50,共3页
Construction Technology
基金
张家口市科学技术研究与发展指导计划项目(1221002B)
