红土抗剪强度指标及其BP网络模型研究

Study on Shear Strength Index of Red Soil and Its BP-Network Model

在分析土体的物理性质指标对其抗剪强度指标的影响的基础上 ,以人工神经网络为基本工具 ,利用其强大的非线性映射能力 ,来预测土体抗剪强度指标 ,为求解其力学参数提供了一条新途径 .结果表明 ,BP网络模型能够充分体现土体物理性质之间的非线性关系 ,所求出的土体抗剪强度指标的预测值与试验值接近 。

Based on the analysis of the relationship between soil physical indexes and shear strength indexes, the artificial neural network with mighty function of non-linear map is used to predict shear strength indexes of soil, which provides a new way to predict the mechanics parameters. It proves to be a good way to use BP-network to map the non-linear relationship among physical indexes. The results of BP-Network Model are close to the ones of the experiment and are more precise than those of the linear regression.