应用前馈神经网络的大柔性机翼阵风响应分析

Aeroelastic Analysis of Large Flexile Wing Based on Machine Learning Algorithm

摘要:

大柔性飞行器因结构重量低、柔性大使得机翼等部件在受载时产生较大的弹性变形，呈现显著的几何非线性效应，因此准确的结构大变形建模方法对于几何非线性气动弹性分析至关重要，而神经网络对非线性系统具有强大的拟合能力，可通过将神经网络应用于非线性结构建模，构造适用于结构大变形的前馈神经网络预测模型，在样本特征和数据结构相对较优的条件下结合曲面涡格法，搭建非线性气动弹性分析框架，对某机翼模型进行阵风响应计算；结果表明神经网络模型能准确预测大柔性机翼结构大变形，应用到气动弹性分析后能进行准确的阵风响应计算，验证了将神经网络应用到结构大变形预测的可行性，为以后机器学习技术与气动弹性分析结合的研究提供思路和方法。

Abstract:

Large flexible spacecraft because of its low weight, flexible, ambassador of the wing structure parts in large elastic deformation occurring during loading, present the geometric nonlinear effect significantly, therefore accurate structure of large deformation is vital for geometric nonlinear aeroelasticity analysis modeling method, and neural network for nonlinear system with strong ability of fitting, By applying neural network to nonlinear structure modeling, the prediction model of large deformation of structure was constructed, and the nonlinear aeroelastic analysis framework was built by combining curved vortex lattice method, and gust response of a wing model was calculated. Results show that the neural network model can accurately predict big flexible wing structure deformation, applied to the pneumatic elastic analysis can accurately gust response calculation, the neural network is applied to the structure is verified the feasibility of large deformation forecast, for the machine learning technology combined with pneumatic elastic analysis of research ideas and methods.