边坡位移预测的神经网络模型研究

STUDY ON NNT MODEL FOR SLOPE DISPLACEMENTS PREDICTION

了解边坡位移的发展趋势对于预测边坡失稳时间具有重要的意义,而导致边坡产生位移的因素是极其复杂的,采用具有非线性映射功能的神经网络模型,运用历史位移数据训练神经网络,并在 Matlab 环境下编程实现,与实测值相比较,神经网络模型具有很高的预测精度.

As slope failure is one of widely distributed geological hazard which treads towards to destroy ecological balance and reduces economic profits,the displacement of slope is basically significant to predict the slope failure.Because of the complication,variety,randomicity,nonstablity of the factors leading to slope displacement,it is very difficult to develop exact explicit models.However,neural network posseses is of strong capacity of nonlinear reflection,and the BP neural network(NNT)model is adopted to forecast the displacements of slope.By means of local measurements in northern slope of Cangshang gold mine,Laizhou,Shandong,the BP NNT is trained.Then the net is tested by comparison between existing measured data and calculated results. Finally the net can be used to calculate the displacements in the future.The whole progress is programmed by Matlab languge.Three conclusions can be found:(1)the programs are easily realized by means of Matlab language,(2)there is tiny difference between calculated result and existing measurements and the model possesses high forecasting accuracy,and(3)with the criterion of slope failure,the model can be used to predict time of slope failure.