人工神经网络技术在现代混凝土配合比设计中的应用

Artificial neural network technology in the modern concrete mix design

各种原材料性能的波动对混凝土性能影响很大,特别对现代混凝土的影响规律很难把握。把水灰比、水泥强度、砂石舍泥量、砂石细度模数、石子的最大粒径、石子针片状含量、掺舍料用量和细度等这些原材料性能参数作为人工神经网络(BP网络)的输入,以对应的优化配合比作为网络的输出,用网络结构描述它们之间的非线性关系。利用数据样本完成网络的训练并进行检验,利用正交试验数据样本训练的BP网络可以预测不同情况下的配合比,预测精度高,完全可以代替繁重的实验室配合比设计。人工神经网络在一定程度上促进了混凝土科学技术的发展。

Huctuations in raw material properties of concrete on a great performance, especially the modem high-performance concrete on the impact of the law is difficult to grasp.The water-cement ratio,the strength of cement,sand and gravel with mud,sand fineness modulus,the maximum size of gravel,stone flake needle content,the amount of mixture and fineness,and other performance parameters of raw materials such as artificial neural network （BP Network） input.Corresponding to optimize the mix as the output of the network,the network structure describe non-linear relationship between them.The use of sample data network to complete the training and testing,the use of orthogonal data sample training BP network can be predicted with the different circumstances than the forecast accuracy,can be hard to replace laboratory mix design.Artificial neural network to a certain extent,promoted the concrete development of science and technology.