基于神经网络的蜂窝夹芯等效弹性参数的预测

针对Gibson公式计算蜂窝夹芯面内等效弹性参数的不足,通过考虑蜂窝夹芯宏观尺度对等效弹性参数的影响,利用贝叶斯正则化神经网络,并结合正交试验和有限元方法,建立了4个主要影响因素与蜂窝夹芯等效弹性参数之间非线性映射关系的神经网络模型,采用BP神经网络成功实现了对蜂窝夹芯等效弹性参数的预测.对于无限宽度蜂窝夹芯,仿真结果与Gibson公式计算结果一致,而对于有限宽度蜂窝夹芯,通过与铝质蜂窝夹芯压缩实验结果进行比较,采用神经网络方法得到的结果与实测结果吻合更好,证实了该方法的有效性和准确性.

Prediction of Equivalent Elastic Parameters of Honeycomb Core Based on Artificial Neural Network

To overcome the drawback of computation of in-plane equivalent elastic parameters based on Gibson formulas for the honeycomb core, an artificial neural network is created to simulate the nonlinear mapping among four primary factors and equivalent elastic parameters of honeycomb core. Then a BP neural network method is adopted to predict those parameters of honeycomb core successfully. The simulation result of infinite width honeycomb core is the same as the result calculated by Gibson formulas, however, it is more approaching the compress test result of finite width aluminum honeycomb core in literature. The accuracy and validity of BP neural network method used in this paper are verified.