摘要
分析了神经网络方法的优缺点 ,然后利用 BP网络的最大误差学习算法对岩土体渗透参数进行反演求解 ,渗流参数决定了流网的形状。根据边坡渗流场的水头观测数据及现场注水试验资料 ,用神经网络方法进行了渗透参数的反分析。以一大型露天采场边坡为例 ,应用该方法确定了边坡岩体的渗透系数 ,并进行了有限元计算 ,编制了相应的计算程序。工程应用表明 ,水头预测值与观测值吻合 ,为边坡的稳定性分析提供了依据。
Advantages and disadvantages of neural network method were analysed.By the maxmium error learning algorithm,inversion solution was made on the infiltration parameters of rock body which decide the flow pattern.Based on the observation data of water head of the slope infiltration flow field and the data of on site water injection test,back analysis of infiltration parameters is made using neural network method.Taking a large open pit slope as the case,this method was used to determine the infiltration coefficient of the rock body of slope.Finit element calculation was made and correspondent computing program compiled.The engineering application has shown that the predicted value of water head coincides with that of observation,providing a basis for the analysis of slope stability.
出处
《矿业快报》
2003年第4期9-11,共3页
Express Information of Mining Industry
关键词
渗流参数
神经网络方法
BP网络
岩土体
边坡
露天采场
稳定性
Neural network
Inversion of parameters
Infiltration flow
Infiltration coefficient of rock body