基于有限元与支持向量机的大坝变形组合预报模型

Combination Forecasting Model of Dam Deformation Based on FEM and SVM

鉴于大坝变形可有效反映大坝的运行状态,而传统统计模型对大坝变形监测数据的拟合和预测并未考虑坝体及坝基物理力学参数的问题,提出了基于有限元与支持向量机(SVM)的大坝变形预报模型,即通过Hypermesh网格划分得到三维有限元模型,导入Abaqus进行静力计算,得到典型水位下的位移分量。考虑到大坝变形的高度非线性特点,利用SVM拟合得到水压分量、温度、时效分量影响过程线。实例验证表明,该模型拟合效果较好,预测精度更高。

Dam deformation can reflect the dam running state effectively,but in the traditional statistical model,the fitting and forecasting of the dam deformation monitoring data do not consider the physical and mechanical parameters of the dam body and dam foundation.Therefore,forecasting model based on FEM and SVM was put forward in this paper.Using Hypermesh to grid partition,three-dimensional finite element model was obtained and imported into Abaqus for static calculation.The displacements were calculated under the typical water level.Considering the highly nonlinearity of the dam deformation,SVM was sued to fit the process line of water pressure component,temperature and aging component.The verification of example shows that the model has better fitting effect and higher prediction precision.