基于神经网络的公路边坡稳定性实时判断

摘要

为了实现对非粘性土公路边坡的稳定性实时预警,采用神经网络方法建立了公路边坡稳定性安全系数Fs和变形值的关系模型,该方法克服了Fs不能实时获取的弊端,由实时测量的变形值计算出Fs,并通过Fs实现无粘性土公路边坡稳定性的实时预警,避免了传统实时预警方法中需要根据经验设定各种变形值阈值的问题.对某无粘性土公路边坡的实验研究表明,神经网络模型计算精度优于其他经验模型,且能够满足工程实时监测的需要.%In order to provide early warning to roadside bank stability problem, This paper builds a mathematical model between safety factors of roadside bank stability Fs and deformation value based on artificial neural network (ANN). It can work out Fs with real-time deformation value to roadside bank, thus overcoming the weakness that Fs can 't be obtained in time. Using this method, real-time warning of cohesionless (non-clay) soil-based roadside bank stability problem can be provided by using Fs. In comparison, for the same purpose the traditional real-time warning methods in roadside bank stability have to set the threshold value of all deformation value. Results from a sample test on a roadside bank with cohesionless soil demonstrate that this model is superior to others in accuracy and adaptability, and it meets the need of real-time monitoring engineering.