摘要
以湿陷性黄土地区的大量强夯工程实例为对象,分析、选择了影响夯沉量的五大主控因素作为BP神经网络模型的基本特征量,建立夯沉量与其之间的相关关系的BP网络模型,对夯沉量进行了预测分析。结果表明:BP神经网络模型能真实反映强夯夯沉量与主控因素之间的非线性关系,预测结果与实测值之间的相对误差小于10%,用该模型对强夯夯沉量进行预测是有效的。
Taking a lot of dynamic compaction projects in collapsible loess area for example, five main factors influencing settlement are analyzed as the basic characteristic quantity of BP neural network model. BP network model of correlation between these factors and settlement is established to predict the settlement. It is indicated BP neural network model can truly reflect the non-linear relationship between the main factors and dynamic compaction settlement. The relative error between the predicted result and measured result is less than 10%. Therefore, this model is effective to predict dynamic compaction settlement.
出处
《路基工程》
2011年第1期67-69,73,共4页
Subgrade Engineering
关键词
强夯
夯沉量
BP神经网络
湿陷性黄土
dynamic compaction
settlement
BP neural network
collapsible loess
