径向基神经网络在粉煤灰混凝土二维和三维碳化分析中的应用

研究了不同粉煤灰掺量(0%,15%,40%,60%)、不同水灰比(0.3,0.35,0.4)、不同强度等级下(C30～C50)高性能混凝土二维和三维碳化深度;应用多因子输入向量的径向基神经网络(RBF)进行二维和三维碳化深度的预测。试验表明,采用多因子输入向量的径向基神经网络能够在试验样本数量较少的情况下建立高效准确的预测网络,可以较好地预测混凝土的碳化深度,其二维和三维碳化深度预测精度比一维精度高,其一维,二维和三维碳化深度的预测值和试测值相对误差分别为10.9%,5.6%,7.1%。混凝土二维和三维碳化研究对混凝土结构耐久性和寿命预测具有现实意义。

Application of RBF Neural Network In analysis of 2 Dimensions and 3 Dimensions Carbonation of Fly Ash Concrete

2 dimensions and 3 dimensions carbonation are studied on different water to cement ratio（0.3,0.35,0.4）,different fly ash proportion（0%,15%,20%,40%,60%） and different strength grade（C30～C50）.Predict the carbonation depth based on RBF neural network using inputting vector of multiple factor.The experiment indicates,2 dimensions and 3 dimensions carbonation can be very good forecast using RBF Neural Network;at the same time,the forecasting network of accurate and high efficiency can be built under fewer experiment sample,and the precision of 2dimensions,3 dimensions is higher than one dimension precision,it＇s relative error of 1 dimension,2 dimensions and 3 dimensions are 10.9%,5.6%,7.1% respectively,between forecasting value and testing.The research of the 2 dimensions and 3 dimensions carbonation of concrete possesses the real meaning to the structure durability and life prediction of concrete.