人工神经网络在水泥加固土力学性能预测中的应用

Application of the principle of artificial neural network in predicting the mechanical properties of cement stabilized soil

提出了基于人工神经网络（ANN）的水泥加固土力学性能指标计算的新方法，并在此基础上预估水泥土搅拌桩体和复合地基的承载力。利用实测资料直接建模，避免了传统方法计算过程中各种人为因素的干扰，所建立的模型预测精度高、简便易行，因而具有广泛的工程实用价值。

Based on artificial neural network, a new method to calculate the mechanical properties of cement-stabilized soil was put forward in this thesis. On the basis of this method, the bearing-capacity of cement-stabilized soil mixing pile body and composite ground can be predicted. The solution takes advantage of in-situ measured data directly for establishing model, thus avoiding the interference of human factors in traditional calculating methods. The model owns high precision and is easy to deal with, thus has extensive practical engineering value.