基于神经网络与杂交算法的机翼结构优化设计

以机翼结构为研究对象,提出了一种能同步解决元件位置优化与尺寸优化的杂交算法.利用MSC.NASTRAN进行尺寸优化,用遗传算法对位置设计变量进行优化,并将尺寸优化结果作为遗传操作的依据,最终实现了尺寸与位置的同步优化.为提高算法效率,利用神经网络的非线性映射功能,对MSC.NASTRAN的尺寸优化结果进行映射以取代其优化过程.算例结果表明,该方法高效、精确,具有很好的推广应用价值.

Optimization design of aerofoil structure based on neural network and hybridization algorithm

Taking the aerofoil structure as the object of study,a kind of hybridization algorithm that could settle synchronistically the optimizations of element's position and of dimensions was put forward.Using MSC.NASTRAN to carrying out dimensional optimization,utilizing genetic algorithm to carrying through optimization on variables of positional design,and take the result of dimensional optimization as the foundation of genetic operation thus finally realized the synchronized optimizations of dimension and position.In order to enhance the efficiency of algorithm,the nonlinear mapping function of neural network was used to carrying out mapping on the result of dimensional optimization of MSC.NASTRAN so as to replace its optimization processes.The result of calculating example showed that the being mentioned method is highly active,accurate and possesses good application value of popularization.