将人工神经网络引入桩基础选型中 ,分析影响桩基础选型的重要因素和 BP神经网络的特点 ,确定网络结构。利用 Matlab 6 .5人工神经网络工具箱构建桩基础选型模型 ,并利用 Matlab语言编制人机交互式界面 ,使选型过程简单明了。将典型工程项目工程地质资料数据作为网络学习训练样本输入建立的神经网络模型 ,对其进行训练、优化。经检验 ,训练好的网络模型性能良好 ,达到了智能预测的目的。

Study on the Selection of Piles with Artificial Neural Network MEI Qing-hai,HUO Kai-cheng(School of Civil Engineering and Architecture, Wuhan University of Technology, Wuhan 430070, China)The artificial neural network was introduced into the selection of piles. The important affecting elements of the selection of piles and the characteristics of BP artificial neural network were analyzed to devise network construction. An identification model of artificial neural network about the selection of piles was established based on the artificial neural network toolbox of Matlab 6.5, and the process of identification was simpler based on human-computer interactive interface by Matlab. The geology data of typical project items was input to the neural network model as practice sample, and then the model was trained and optimized. Through testing, the trained model can help designers to select an appropriate kind of piles.BP neural network; Matlab; selection of piles; training network