基于人工神经网络的涡轮盘蠕变可靠性分析方法

Artificial Neural Networks-Based Creep Reliability Analysis Method of Turbine Disk

用ADINA程序计算的涡轮盘盘缘蠕变位移量的有限个结果对人工神经网络进行训练。利用Monte-Carlo法对基本随机变量进行随机抽样,并基于随机有限元方法思想用人工神经网络代替有限元程序计算涡轮盘盘缘蠕变位移量,进而分析涡轮盘蠕变可靠性。

The basic variables are sampled by MonteCarlo method. Instead of complicate finite element analysis,a Neural Network,which was well trained by the results of ADINA program,is used to calculate the displacement at the rim of turbine disc.As a result,this paper shows how to use Neural Network method to calculate the creep reliability of an aeroengine turbine disc.