利用 ANSYS 进行海堤的非稳定渗流的有限元分析, 并提取渗压计测点的时序渗压和对应的渗透系数分别作 为神经网络的输入层和输出层, 建立海堤渗压与渗透系数的非线性映射关系。在此基础上, 将实测渗压序列代入训 练好的 BP 神经网络中进行海堤渗透系数的反演, 再将反演成果投入有限元正分析, 结果显示, 测点渗压的模拟计 算值与实测渗压序列的大小及变化趋势相同, 说明拟合效果理想。

ANSYS was used to perform the finite element analysis of unsteady seepage of seawall. T he osmotic pressure sequence and corresponding permeability parameters were collected to act as the input layer and output layer of the neural network, and the nonlinear mapping relationship between the osmotic pressure and permeability parameters of seawall was developed. Based on the results, the observed osmotic pressure sequence was put into the trained BP neural network to conduct the back analysis of the permeability parameters of seawall. The inversion results were analyzed using the finite element method, which showed that the simulated and measured osmotic pressures at the monitoring points are similar.

国家自然科学基金项目/ 特殊工作环境下海堤堤身状态的实时监测分析及预警研究0( 50979056)