隐层自构造小波神经网络及其在深基坑支护工程中的应用

在结合小波分析和神经网络（ANN）基础上，提出了一种隐层自构造小波神经网络模型。该模型通过有限的经验数据学习，建立了深基坑支护结构变形量与其影响因素的非线性关系，并将其应用于深基坑支护工程实例当中。研究表明，该网络训练时间不到0．5s，预测精度高，预测结果可靠，对今后保证深基坑支护工程施工安全具有借鉴意义。

Middle layer self-construction of wavelet neural network and its application in engineering of support of deep foundation pit

On the basis of combining wavelet analysis and neural network,a model of hidden layer self-construction wavelet of neural network was proposed. The nonlinear relation between deformation quantity of deep foundation pit and its influencing factors was obtained from the finite empirical data by the model, which was applied to the engineering of supporting structure of deep foundation pit. The research shows that the net training time is less than 0.5 s,the predicting accuracy is high and the predicting result is reliable, which possesses reference meaning in the guarantee of construction security in future.