基于反演的LS-SVM及其在土石坝沉降中的应用

Application of LS-SVM Based on Back-analysis in Rockfill Dam Settlement Analysis

仿效BP神经网络在反演分析中的应用,利用神经网络模拟LS-SVM参数与计算结果间的关系,提出了反演参数的方法,并将其应用于陆浑水库沉降预测中。通过与神经网络的沉降预测值比较,表明该方法具有较好的可行性以及较高的预测精度,适合于解决土石坝沉降预测问题。

As like as the application of BP Neural Network in back-analysis,the method that back analyses the two parameter has been introduced by simulating the relation of LS-SVM＇s parameter and its result by BP Neural Network. In comparison with BP Neural Network in the prediction of the Luhun reservoir＇s settlement,the LS-SVM algorithm has good feasibility and efficiency, and possesses the higher forecasting accuracy,and can solve the settlement prediction of the rockfill dam.