论文部分内容阅读
基于人工神经网络方法 ,根据云峰大坝坝顶水平位移观测资料能够识别大坝混凝土和岩石基础的弹性模量 .采用 BP学习算法 ,并通过增加阻尼项以及对观测数据的归一化处理 ,避免了迭代过程中的振荡性 ,提高了参数识别精度 .将弹性模量识别结果代入到有限元模型中 ,计算所得到的坝顶水平位移与坝顶观测水平位移水压分量的最大误差小于 0 .15mm.工程实际应用表明 ,用神经网络方法识别材料参数具有识别精度高和收敛速度快等特性 .
Based on the artificial neural network method, the elastic modulus of dam concrete and rock foundation can be identified according to the horizontal displacement observation data of dam peak in Yunfeng Dam.Based on the BP learning algorithm and by adding the damping term and the normalized observation data, Avoiding the oscillation in the iteration process and improving the accuracy of parameter identification.The elastic modulus identification result is substituted into the finite element model and the maximum error between the calculated horizontal displacement of crest and the hydraulic component of horizontal displacement at the crest is less than 0 .15mm. The practical application shows that the neural network method to identify the material parameters with high recognition accuracy and fast convergence and other characteristics.