一种高效高精度的气动弹性结构优化方法

气动弹性结构优化技术主要包括约束求解和优化算法两个方面的内容。针对常用的基于低阶面元法的静气动弹性分析方法计算效率高但精度低的特点，建立了一种高效高精度的基于高阶面元法的静气动弹性分析方法。针对当前气动弹性结构优化技术使用单一优化算法导致搜索精度低、收敛速度慢等特点，将遗传算法和分形算法进行结合，发展了一种遗传/分形混合算法。针对气动弹性结构优化计算时间长、设备要求高等特点，引入了Kriging代理模型方法来加快优化速度，减少时间和设备的耗费。最后以某大展弦比客机机翼为算例，采用基于高阶面元法的静气动弹性分析方法求解约束响应样本，用Kriging代理模型方法对约束响应进行模型构建和预测，并将Kriging代理模型和遗传/分形混合优化算法进行结合，构建了一种高效高精度的静气动弹性结构优化方法。优化分析结果表明，Kriging代理模型在静气动弹性响应预测上具有很高的精度，平均误差均在5%以下，副翼效率预测的平均误差甚至低于1%；遗传/分形混合算法相比于单一的遗传算法具有更快的收敛速度和更强的全局寻优能力。

A High Efficient and Accurate Method for Aeroelastic Structure Optimization

Aeroelastic structure optimization technique mainly consists of two aspects: constraint solving and optimization algorithms. Aiming at the high efficiency but low accuracy of static aeroelasticity analysis method based on lower order panel method, a high precision static aeroelasticity analysis method based on high order panel method is established. Aiming at the characteristics of the current aeroelastic structure optimization technology that the single optimization algorithm results in low search accuracy and slow convergence speed, a genetic/fractal hybrid algorithm is developed by combining genetic algorithm and fractal algorithm. Aiming at the characteristics of long computational time and high device requirement of static aeroelasticity structure optimization, the Kriging surrogate model is introduced to speed up optimization and reduce the time and device cost. Finally, taking a large aspect ratio aircraft wing as an example, the static aeroelastic analysis method based on higher order panel method is used to solve the constrained response samples, and the Kriging surrogate model method is used to construct and predict the constraint response. Combined the Kriging surrogate model with the genetic/fractal hybrid optimization algorithm, a high efficiency and high precision static aeroelastic structure optimization method is built. The result of optimization analysis shows that the Kriging surrogate model has high precision in predicting the static aeroelastic response with the average error below 5%, and the average error of the aileron efficiency forecast even below 1%. Compared to the single genetic algorithm, the genetic/fractal hybrid algorithm has faster convergence speed and stronger global optimization ability.