基于改进BP神经网络的孔板应力集中系数预测

Stress Concentration Factors Prediction of Orifice Plate Based on Modified BP Neural Network

在正交试验和无网格法数值计算结果的基础上,针对常用BP算法的不足,采用动量因子与自适应学习速率相结合的改进BP神经网络方法,建立了孔板应力集中系数预测模型。经过计算结果的检验,表明该模型是可行的,对今后孔板应力集中系数预测具有借鉴意义。

Based on orthogonal experiment and numerical calculation results of meshless method, in light of deficiency of the normal BP algorithm, a neural network predictive model for the stress concentration factors of orifice plates established on the basis of the improved BP algorithm in combination of momentum factor and self- adaptive learning rate. Tested by results which export from computer, this model is proved to be feasible, and offers reference to the stress concentration factor prediction of orifice plate.