利用GRNN方法分析地下管道抗震性能影响因素的敏感性

Parameter sensitivity analysis of seismic response of underground pipeline based on GRNN

通过对地震动作用下地下管道最大轴向应力和多种影响因素的分析,基于广义回归神经网络(GRNN)建立了地下管道地震反应预测模型。在此基础之上,将正交表试验设计理论、效用函数理论与神经网络结合起来对地下管道抗震性能影响因素进行敏感性分析,并给出了敏感性分析的具体算法。最后,通过实例分析,将分析结果与用传统理论分析所得结果进行比较,结果显示,该方法不仅可靠,而且简单方便。

Through the analysis of the maximum axial stress under the action of seismic ground motion and the affecting factors, the prediction model of underground pipeline response under earthquakes was constructed by using GRNN. On the basis of the model, combining the orthogonal array with the theory of utility function and ANN, the assessment model was set up to make sensibility analysis. The authors made a comparison between the results obtained from the traditional limit balance analysis and the results from the proposed method. It is shown that the results from the proposed method are valid and convenient.