摘要
黄土高边坡土体的稳定主要受坡体含水量的控制,如果能提前根据含水量的变化预测边坡的安全系数和安全状态,就能及时采取安全措施,减少或避免由边坡失稳造成的损失。在搜集黄土高边坡工程典型实例资料的基础上,综合考虑影响边坡稳定的因素,根据边坡的几何、物理、力学参数构建训练样本和测试样本,基于LM算法的BP神经网络建立了黄土高边坡稳定性预测模型,并由资料拟合出黄土强度参数与含水量的关系式,由此提出了利用观测含水量预测关中地区高边坡稳定性的系统模型。最后以关中地区某一高边坡为例,简要介绍了该模型的使用。结果表明,模型的预测值和期望值吻合较好,具有较高的可靠性;当含水量超过13.4%时该边坡失稳,与实际情况吻合,说明该模型在关中地区具有较强的实用性。
Water content is the major factor influencing the stability of the high loess slopes. If the safety coefficieut and safety state of the slope can be predicted based on the changes of water content ahead of time,peope will take effective and active measures in time,then the loss caused by the slope instability can be decreased and even avoided. This paper used numerous cases of high loess slopes, considering the influencing factors, formed the data sets for training and testing based on the geometry, physics, mechanics parameters of the slopes, and established prediction model to estimate stability of high loess slope by BP neural network based on Levenberg-Marquardt Algorithm, then fitted the relationship between strength parameters and water content of loess by experimental data ,and presented the system model to predict the stability of high loess slopes by measuring water content, finally took one Loess slope in Guanzhong area for example, briefly introduced the application of system model. The result showed that the slope would be instable when the water content reached more than 13.4% ,which was consistent with the practical situation. So,it＇s proved that this model had strong practicability with high precision in Guanzhong area.
出处
《西北农林科技大学学报（自然科学版）》
CSCD
北大核心
2007年第6期202-206,共5页
Journal of Northwest A&F University(Natural Science Edition)
关键词
含水量
黄土高边坡
稳定性预测模型
LM算法
BP神经网络
water content
high loess slope
prediction model to estimate stability
LM algorithm
BP neural network
