基于神经网络的垂尾飞行载荷模型研究

摘要

Aiming at the complexity of structural design of modern aircraft.multiple paths of strength transfer,large structural deformation under big loads and so on.the method using the neural networks to build loads model is proposed.Utilizing the calibrated testing data of loads of vertical tail,the loads model of shear and bending moment of vertical tail's root based on neural networks is founded.Through comparing and analyzing the results of loads prediction between multiple points linear regression.we find that the prediction under big loads exceed the result by multiple linear regression.So the method is reliable for measuring the flight loads of vertical tail.%针对现代飞机结构设计复杂、传力路径多和大载荷结构变形大等因素,提出用神经网络方法建立载荷模型.利用垂尾载荷校准试验数据建立基于神经网络的垂尾根部剪力和弯矩的载荷模型,并对多点进行载荷预测.通过与多元线性回归载荷预测结果进行对比分析,结果表明,在大载荷下神经网络预测精度优于多元线性回归,从而为测量垂尾飞行载荷提供了一种可靠的方法.