混凝土强度预测与设计的神经网络方法

首次提出混凝土强度预测与设计的神经网络方法。对普通混凝土、高强混凝土等多种混凝土建立了表现混凝土强度因果规律的神经网络方法，在已有混凝土试验资料基础上，进行了强度预测。与通用的回归分析方法所得结果对比，神经网络方法精度较高。通过实例给出了应用神经网络方法的强度预测结果。表明神经网络是混凝土强度，特别是高强混凝土强度预测和设计的崭新途径

Neural Network Method for Concrete Strength Prediction and Design

Presented are researches on strength prediction of common concrete, fly ash concrete, and high strength fly ash silica fume concrete by artificial neural network (ANN) method. The maximum prediction errors of learned and unlearned samples are less than 5%. It is concluded that ANN is an absolutely new concrete strength prediction method, particularly suitable for strength prediction of high strength concrete.