基于人工神经网络的结构零件可靠度仿真方法

SIMULATION METHOD FOR STRUCTURE MEMBER RELIABILITY BASED ON ARTIFICIAL NEURAL NETWORKS

提出结构零件可靠度计算的人工神经网络—蒙特卡罗仿真算法。针对结构零件可靠度计算问题，利用有限元计算方法，得到一组结构输入基本变量及输出响应数据，采用人工神经网络后向传播ＢＰ模型，建立结构的输入—输出关系，在该模型上，用Ｍｏｎｔｅ Ｃａｒｌｏ 方法进行结构零件可靠度仿真计算。算例表明。

In this paper, an artificial neural networks Monte Carlo simulation method is presented for structure member reliability calculation. For the structure member, a group of data of basic input variables and output responses can be obtained by finite element method. Then the back propagation feed forward BP artificial neural networks are used for modeling the relation of structural input and output. Using this model, the structure member reliability simulation calculation can carry out by Monte Carlo method. Examples show that this method is available and feasible.