基于复合形与神经网络的机翼结构优化设计

Structure design optimization of the wing based on neural networks and complex method

从元件位置和元件尺寸两个方面考虑,提出了一种全局结构优化设计方法。利用NASTRAN完成尺寸优化,利用神经网络对NASTRAN优化结果进行映射,用复合形法进行位置寻优,最终完成了机翼结构的优化。算例表明利用神经网络来预测NASTRAN优化结果有较高的精度与适用性,本文所提方法可行、正确,具有很高的效率。

From the thinking of both position variables and size variables, a global method that can solve the optimization problems was present in this paper. Letting the NASTRAN optimize the size variables, using the neural network to predict the calculated result of the NASTRAN, and this paper went on to optimize the position variables with the complex method. The example shows the structural optimization method that use Neural Networks to predict the result of the NASTRAN has high precision and adaptability, this method is not only correct and feasible, but highly effective and practically convenient.